Upload checkpoint-25000

.gitattributes

@@ -38,3 +38,4 @@ checkpoint-15000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 checkpoint-5000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 checkpoint-20000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-25000/tokenizer.json filter=lfs diff=lfs merge=lfs -text

checkpoint-25000/chat_template.jinja

@@ -0,0 +1,93 @@
+{{- bos_token }}
+{%- if custom_tools is defined %}
+    {%- set tools = custom_tools %}
+{%- endif %}
+{%- if not tools_in_user_message is defined %}
+    {%- set tools_in_user_message = true %}
+{%- endif %}
+{%- if not date_string is defined %}
+    {%- if strftime_now is defined %}
+        {%- set date_string = strftime_now("%d %b %Y") %}
+    {%- else %}
+        {%- set date_string = "26 Jul 2024" %}
+    {%- endif %}
+{%- endif %}
+{%- if not tools is defined %}
+    {%- set tools = none %}
+{%- endif %}
+
+{#- This block extracts the system message, so we can slot it into the right place. #}
+{%- if messages[0]['role'] == 'system' %}
+    {%- set system_message = messages[0]['content']|trim %}
+    {%- set messages = messages[1:] %}
+{%- else %}
+    {%- set system_message = "" %}
+{%- endif %}
+
+{#- System message #}
+{{- "<|start_header_id|>system<|end_header_id|>\n\n" }}
+{%- if tools is not none %}
+    {{- "Environment: ipython\n" }}
+{%- endif %}
+{{- "Cutting Knowledge Date: December 2023\n" }}
+{{- "Today Date: " + date_string + "\n\n" }}
+{%- if tools is not none and not tools_in_user_message %}
+    {{- "You have access to the following functions. To call a function, please respond with JSON for a function call." }}
+    {{- 'Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}.' }}
+    {{- "Do not use variables.\n\n" }}
+    {%- for t in tools %}
+        {{- t | tojson(indent=4) }}
+        {{- "\n\n" }}
+    {%- endfor %}
+{%- endif %}
+{{- system_message }}
+{{- "<|eot_id|>" }}
+
+{#- Custom tools are passed in a user message with some extra guidance #}
+{%- if tools_in_user_message and not tools is none %}
+    {#- Extract the first user message so we can plug it in here #}
+    {%- if messages | length != 0 %}
+        {%- set first_user_message = messages[0]['content']|trim %}
+        {%- set messages = messages[1:] %}
+    {%- else %}
+        {{- raise_exception("Cannot put tools in the first user message when there's no first user message!") }}
+{%- endif %}
+    {{- '<|start_header_id|>user<|end_header_id|>\n\n' -}}
+    {{- "Given the following functions, please respond with a JSON for a function call " }}
+    {{- "with its proper arguments that best answers the given prompt.\n\n" }}
+    {{- 'Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}.' }}
+    {{- "Do not use variables.\n\n" }}
+    {%- for t in tools %}
+        {{- t | tojson(indent=4) }}
+        {{- "\n\n" }}
+    {%- endfor %}
+    {{- first_user_message + "<|eot_id|>"}}
+{%- endif %}
+
+{%- for message in messages %}
+    {%- if not (message.role == 'ipython' or message.role == 'tool' or 'tool_calls' in message) %}
+        {{- '<|start_header_id|>' + message['role'] + '<|end_header_id|>\n\n'+ message['content'] | trim + '<|eot_id|>' }}
+    {%- elif 'tool_calls' in message %}
+        {%- if not message.tool_calls|length == 1 %}
+            {{- raise_exception("This model only supports single tool-calls at once!") }}
+        {%- endif %}
+        {%- set tool_call = message.tool_calls[0].function %}
+        {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' -}}
+        {{- '{"name": "' + tool_call.name + '", ' }}
+        {{- '"parameters": ' }}
+        {{- tool_call.arguments | tojson }}
+        {{- "}" }}
+        {{- "<|eot_id|>" }}
+    {%- elif message.role == "tool" or message.role == "ipython" %}
+        {{- "<|start_header_id|>ipython<|end_header_id|>\n\n" }}
+        {%- if message.content is mapping or message.content is iterable %}
+            {{- message.content | tojson }}
+        {%- else %}
+            {{- message.content }}
+        {%- endif %}
+        {{- "<|eot_id|>" }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' }}
+{%- endif %}

checkpoint-25000/config.json

@@ -0,0 +1,36 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "dtype": "bfloat16",
+  "eos_token_id": 128009,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 3072,
+  "initializer_range": 0.02,
+  "intermediate_size": 8192,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 24,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": 128263,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 32.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": true,
+  "transformers_version": "4.57.1",
+  "use_cache": false,
+  "vocab_size": 156960
+}

checkpoint-25000/generation_config.json

@@ -0,0 +1,13 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 128000,
+  "do_sample": true,
+  "eos_token_id": [
+    128009,
+    128258
+  ],
+  "pad_token_id": 128263,
+  "temperature": 0.6,
+  "top_p": 0.9,
+  "transformers_version": "4.57.1"
+}

checkpoint-25000/model-00001-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1dae409f70c5beb92916662c6bc389b9b235ac8aa5edd19a4dcb87e37a73074
+size 4991160848

checkpoint-25000/model-00002-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df22e9a90c1bea262250982640b119e6020474736991da482cb6ed56dd23d045
+size 1610725592

checkpoint-25000/model.safetensors.index.json

@@ -0,0 +1,262 @@
+{
+  "metadata": {
+    "total_parameters": 3300928512,
+    "total_size": 6601857024
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.norm.weight": "model-00002-of-00002.safetensors"
+  }
+}

checkpoint-25000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7394351b5e12259584bf51e2b8b1838561912dfc12d16a418e4a2e9977d9f005
+size 13203939671

checkpoint-25000/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fd22a327165f740f1eab25c0d978e67e1cd82aa9b8724fdd843408620c72ff3a
+size 16389

checkpoint-25000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5673161836506c384e0abc8477bff3b0855a6d2774d863ccdf3826b2aeefbb60
+size 16389

checkpoint-25000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bb4614864db57508789a0df926c228e0d11dba5c98649f5829eac6c4188218c
+size 16389

checkpoint-25000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0db1d570882d273319727563cceae81d94a17b383fca92892fae4e135d2775ad
+size 16389

checkpoint-25000/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb89ba586002a8ac9b5b1e468fa891f3dc829f56b8a7d09d57ae7b16a5f8bced
+size 16389

checkpoint-25000/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4a3382152c2a3ccf97bffb7c42bfbf02c5fa050670dac34327da8cf6e6bdd283
+size 16389

checkpoint-25000/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:650539683ae2330d8bc96f4df5e6b0624b4b92ff108f322857f936eba5b38828
+size 16389

checkpoint-25000/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:313d47ba251ff35c35ebefbb2ccc15d46ea4977fa09bc738f1c1d161806b5274
+size 16389

checkpoint-25000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:58028f339faecd4c4b7e1ffd636533e8278a206434651a70ee0bb05022d2698e
+size 1465

checkpoint-25000/special_tokens_map.json

@@ -0,0 +1,165 @@
+{
+  "additional_special_tokens": [
+    {
+      "content": "<angry>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<appalled>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<chuckle>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<cry>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<curious>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<disappointed>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<excited>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<exhale>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<gasp>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<giggle>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<gulp>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<laugh>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<laugh_harder>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<mischievous>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<sarcastic>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<scream>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<sigh>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<sing>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<snort>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<whisper>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false
+    }
+  ],
+  "bos_token": {
+    "content": "<|begin_of_text|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|eot_id|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<custom_token_7>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}

checkpoint-25000/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c5e5b1d89b7e3738e5a5a4f93c326d8f3292ea83f9c560b8dbb6d66fb851973
+size 22853258

checkpoint-25000/trainer_state.json

@@ -0,0 +1,3581 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 8.283655041007373,
+  "eval_steps": 5000,
+  "global_step": 25000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0003313727114572115,
+      "grad_norm": 2.828125,
+      "learning_rate": 0.0,
+      "loss": 4.1855,
+      "step": 1
+    },
+    {
+      "epoch": 0.016568635572860577,
+      "grad_norm": 3.234375,
+      "learning_rate": 9.017298490982701e-07,
+      "loss": 4.4012,
+      "step": 50
+    },
+    {
+      "epoch": 0.03313727114572115,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.8218623481781377e-06,
+      "loss": 4.3839,
+      "step": 100
+    },
+    {
+      "epoch": 0.04970590671858172,
+      "grad_norm": 2.265625,
+      "learning_rate": 2.741994847258005e-06,
+      "loss": 4.345,
+      "step": 150
+    },
+    {
+      "epoch": 0.0662745422914423,
+      "grad_norm": 2.171875,
+      "learning_rate": 3.662127346337873e-06,
+      "loss": 4.2973,
+      "step": 200
+    },
+    {
+      "epoch": 0.08284317786430287,
+      "grad_norm": 1.5234375,
+      "learning_rate": 4.582259845417741e-06,
+      "loss": 4.224,
+      "step": 250
+    },
+    {
+      "epoch": 0.09941181343716345,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.502392344497608e-06,
+      "loss": 4.144,
+      "step": 300
+    },
+    {
+      "epoch": 0.11598044901002402,
+      "grad_norm": 1.0390625,
+      "learning_rate": 6.422524843577475e-06,
+      "loss": 4.0929,
+      "step": 350
+    },
+    {
+      "epoch": 0.1325490845828846,
+      "grad_norm": 1.1796875,
+      "learning_rate": 7.342657342657343e-06,
+      "loss": 4.0516,
+      "step": 400
+    },
+    {
+      "epoch": 0.14911772015574518,
+      "grad_norm": 1.390625,
+      "learning_rate": 8.26278984173721e-06,
+      "loss": 4.025,
+      "step": 450
+    },
+    {
+      "epoch": 0.16568635572860574,
+      "grad_norm": 1.515625,
+      "learning_rate": 9.182922340817078e-06,
+      "loss": 4.0113,
+      "step": 500
+    },
+    {
+      "epoch": 0.18225499130146633,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.0103054839896946e-05,
+      "loss": 4.0015,
+      "step": 550
+    },
+    {
+      "epoch": 0.1988236268743269,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.1023187338976813e-05,
+      "loss": 3.9936,
+      "step": 600
+    },
+    {
+      "epoch": 0.21539226244718748,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.1943319838056682e-05,
+      "loss": 3.9778,
+      "step": 650
+    },
+    {
+      "epoch": 0.23196089802004805,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.2863452337136547e-05,
+      "loss": 3.971,
+      "step": 700
+    },
+    {
+      "epoch": 0.24852953359290864,
+      "grad_norm": 1.2265625,
+      "learning_rate": 1.3783584836216415e-05,
+      "loss": 3.9664,
+      "step": 750
+    },
+    {
+      "epoch": 0.2650981691657692,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.4703717335296282e-05,
+      "loss": 3.9561,
+      "step": 800
+    },
+    {
+      "epoch": 0.28166680473862976,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.562384983437615e-05,
+      "loss": 3.9459,
+      "step": 850
+    },
+    {
+      "epoch": 0.29823544031149035,
+      "grad_norm": 1.0,
+      "learning_rate": 1.6543982333456018e-05,
+      "loss": 3.9438,
+      "step": 900
+    },
+    {
+      "epoch": 0.31480407588435094,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.7464114832535886e-05,
+      "loss": 3.9386,
+      "step": 950
+    },
+    {
+      "epoch": 0.3313727114572115,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.8384247331615755e-05,
+      "loss": 3.9285,
+      "step": 1000
+    },
+    {
+      "epoch": 0.34794134703007207,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.930437983069562e-05,
+      "loss": 3.9211,
+      "step": 1050
+    },
+    {
+      "epoch": 0.36450998260293266,
+      "grad_norm": 1.0859375,
+      "learning_rate": 2.022451232977549e-05,
+      "loss": 3.9216,
+      "step": 1100
+    },
+    {
+      "epoch": 0.38107861817579325,
+      "grad_norm": 0.9921875,
+      "learning_rate": 2.1144644828855357e-05,
+      "loss": 3.9118,
+      "step": 1150
+    },
+    {
+      "epoch": 0.3976472537486538,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.2064777327935222e-05,
+      "loss": 3.9059,
+      "step": 1200
+    },
+    {
+      "epoch": 0.4142158893215144,
+      "grad_norm": 1.03125,
+      "learning_rate": 2.298490982701509e-05,
+      "loss": 3.9019,
+      "step": 1250
+    },
+    {
+      "epoch": 0.43078452489437496,
+      "grad_norm": 1.0234375,
+      "learning_rate": 2.390504232609496e-05,
+      "loss": 3.9035,
+      "step": 1300
+    },
+    {
+      "epoch": 0.4473531604672355,
+      "grad_norm": 0.8671875,
+      "learning_rate": 2.4825174825174828e-05,
+      "loss": 3.8937,
+      "step": 1350
+    },
+    {
+      "epoch": 0.4639217960400961,
+      "grad_norm": 0.89453125,
+      "learning_rate": 2.5745307324254693e-05,
+      "loss": 3.881,
+      "step": 1400
+    },
+    {
+      "epoch": 0.4804904316129567,
+      "grad_norm": 0.90234375,
+      "learning_rate": 2.666543982333456e-05,
+      "loss": 3.8808,
+      "step": 1450
+    },
+    {
+      "epoch": 0.49705906718581727,
+      "grad_norm": 0.84375,
+      "learning_rate": 2.7585572322414427e-05,
+      "loss": 3.8782,
+      "step": 1500
+    },
+    {
+      "epoch": 0.5136277027586779,
+      "grad_norm": 1.0,
+      "learning_rate": 2.8505704821494296e-05,
+      "loss": 3.8817,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5301963383315385,
+      "grad_norm": 0.94140625,
+      "learning_rate": 2.942583732057416e-05,
+      "loss": 3.8718,
+      "step": 1600
+    },
+    {
+      "epoch": 0.5467649739043989,
+      "grad_norm": 0.80859375,
+      "learning_rate": 3.034596981965403e-05,
+      "loss": 3.8675,
+      "step": 1650
+    },
+    {
+      "epoch": 0.5633336094772595,
+      "grad_norm": 0.75390625,
+      "learning_rate": 3.12661023187339e-05,
+      "loss": 3.8699,
+      "step": 1700
+    },
+    {
+      "epoch": 0.5799022450501201,
+      "grad_norm": 0.77734375,
+      "learning_rate": 3.2186234817813766e-05,
+      "loss": 3.8645,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5964708806229807,
+      "grad_norm": 0.86328125,
+      "learning_rate": 3.3106367316893635e-05,
+      "loss": 3.8601,
+      "step": 1800
+    },
+    {
+      "epoch": 0.6130395161958413,
+      "grad_norm": 0.85546875,
+      "learning_rate": 3.4026499815973504e-05,
+      "loss": 3.8513,
+      "step": 1850
+    },
+    {
+      "epoch": 0.6296081517687019,
+      "grad_norm": 0.828125,
+      "learning_rate": 3.4946632315053365e-05,
+      "loss": 3.8583,
+      "step": 1900
+    },
+    {
+      "epoch": 0.6461767873415625,
+      "grad_norm": 0.8359375,
+      "learning_rate": 3.5866764814133234e-05,
+      "loss": 3.857,
+      "step": 1950
+    },
+    {
+      "epoch": 0.662745422914423,
+      "grad_norm": 0.796875,
+      "learning_rate": 3.67868973132131e-05,
+      "loss": 3.8488,
+      "step": 2000
+    },
+    {
+      "epoch": 0.6793140584872835,
+      "grad_norm": 0.7734375,
+      "learning_rate": 3.770702981229297e-05,
+      "loss": 3.843,
+      "step": 2050
+    },
+    {
+      "epoch": 0.6958826940601441,
+      "grad_norm": 0.79296875,
+      "learning_rate": 3.862716231137284e-05,
+      "loss": 3.8413,
+      "step": 2100
+    },
+    {
+      "epoch": 0.7124513296330047,
+      "grad_norm": 0.8046875,
+      "learning_rate": 3.954729481045271e-05,
+      "loss": 3.836,
+      "step": 2150
+    },
+    {
+      "epoch": 0.7290199652058653,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.046742730953258e-05,
+      "loss": 3.8366,
+      "step": 2200
+    },
+    {
+      "epoch": 0.7455886007787259,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.138755980861244e-05,
+      "loss": 3.8347,
+      "step": 2250
+    },
+    {
+      "epoch": 0.7621572363515865,
+      "grad_norm": 0.80078125,
+      "learning_rate": 4.230769230769231e-05,
+      "loss": 3.8299,
+      "step": 2300
+    },
+    {
+      "epoch": 0.778725871924447,
+      "grad_norm": 0.75390625,
+      "learning_rate": 4.3227824806772176e-05,
+      "loss": 3.8187,
+      "step": 2350
+    },
+    {
+      "epoch": 0.7952945074973076,
+      "grad_norm": 0.76953125,
+      "learning_rate": 4.4147957305852044e-05,
+      "loss": 3.8225,
+      "step": 2400
+    },
+    {
+      "epoch": 0.8118631430701682,
+      "grad_norm": 0.81640625,
+      "learning_rate": 4.506808980493191e-05,
+      "loss": 3.8229,
+      "step": 2450
+    },
+    {
+      "epoch": 0.8284317786430287,
+      "grad_norm": 0.77734375,
+      "learning_rate": 4.598822230401178e-05,
+      "loss": 3.8245,
+      "step": 2500
+    },
+    {
+      "epoch": 0.8450004142158893,
+      "grad_norm": 0.78515625,
+      "learning_rate": 4.690835480309165e-05,
+      "loss": 3.8109,
+      "step": 2550
+    },
+    {
+      "epoch": 0.8615690497887499,
+      "grad_norm": 0.78125,
+      "learning_rate": 4.782848730217152e-05,
+      "loss": 3.8195,
+      "step": 2600
+    },
+    {
+      "epoch": 0.8781376853616105,
+      "grad_norm": 0.7890625,
+      "learning_rate": 4.874861980125138e-05,
+      "loss": 3.8125,
+      "step": 2650
+    },
+    {
+      "epoch": 0.894706320934471,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.966875230033125e-05,
+      "loss": 3.8131,
+      "step": 2700
+    },
+    {
+      "epoch": 0.9112749565073316,
+      "grad_norm": 0.73046875,
+      "learning_rate": 4.999998362078322e-05,
+      "loss": 3.819,
+      "step": 2750
+    },
+    {
+      "epoch": 0.9278435920801922,
+      "grad_norm": 0.75,
+      "learning_rate": 4.999989244747393e-05,
+      "loss": 3.8082,
+      "step": 2800
+    },
+    {
+      "epoch": 0.9444122276530528,
+      "grad_norm": 0.79296875,
+      "learning_rate": 4.9999721297876855e-05,
+      "loss": 3.8146,
+      "step": 2850
+    },
+    {
+      "epoch": 0.9609808632259134,
+      "grad_norm": 0.765625,
+      "learning_rate": 4.999947017253951e-05,
+      "loss": 3.8042,
+      "step": 2900
+    },
+    {
+      "epoch": 0.977549498798774,
+      "grad_norm": 0.734375,
+      "learning_rate": 4.9999139072265274e-05,
+      "loss": 3.8072,
+      "step": 2950
+    },
+    {
+      "epoch": 0.9941181343716345,
+      "grad_norm": 0.7578125,
+      "learning_rate": 4.9998727998113335e-05,
+      "loss": 3.8008,
+      "step": 3000
+    },
+    {
+      "epoch": 1.0106039267666307,
+      "grad_norm": 0.31640625,
+      "learning_rate": 4.999823695139877e-05,
+      "loss": 3.7924,
+      "step": 3050
+    },
+    {
+      "epoch": 1.0271725623394914,
+      "grad_norm": 0.306640625,
+      "learning_rate": 4.999766593369246e-05,
+      "loss": 3.7963,
+      "step": 3100
+    },
+    {
+      "epoch": 1.0437411979123519,
+      "grad_norm": 0.33203125,
+      "learning_rate": 4.999701494682112e-05,
+      "loss": 3.7837,
+      "step": 3150
+    },
+    {
+      "epoch": 1.0603098334852126,
+      "grad_norm": 0.298828125,
+      "learning_rate": 4.999628399286731e-05,
+      "loss": 3.7942,
+      "step": 3200
+    },
+    {
+      "epoch": 1.076878469058073,
+      "grad_norm": 0.298828125,
+      "learning_rate": 4.99954730741694e-05,
+      "loss": 3.7819,
+      "step": 3250
+    },
+    {
+      "epoch": 1.0934471046309335,
+      "grad_norm": 0.296875,
+      "learning_rate": 4.999458219332157e-05,
+      "loss": 3.7868,
+      "step": 3300
+    },
+    {
+      "epoch": 1.1100157402037942,
+      "grad_norm": 0.294921875,
+      "learning_rate": 4.9993611353173794e-05,
+      "loss": 3.7924,
+      "step": 3350
+    },
+    {
+      "epoch": 1.1265843757766547,
+      "grad_norm": 0.29296875,
+      "learning_rate": 4.999256055683187e-05,
+      "loss": 3.7884,
+      "step": 3400
+    },
+    {
+      "epoch": 1.1431530113495154,
+      "grad_norm": 0.287109375,
+      "learning_rate": 4.999142980765736e-05,
+      "loss": 3.7875,
+      "step": 3450
+    },
+    {
+      "epoch": 1.159721646922376,
+      "grad_norm": 0.3125,
+      "learning_rate": 4.9990219109267596e-05,
+      "loss": 3.7827,
+      "step": 3500
+    },
+    {
+      "epoch": 1.1762902824952366,
+      "grad_norm": 0.314453125,
+      "learning_rate": 4.9988928465535686e-05,
+      "loss": 3.7832,
+      "step": 3550
+    },
+    {
+      "epoch": 1.192858918068097,
+      "grad_norm": 0.28515625,
+      "learning_rate": 4.9987557880590486e-05,
+      "loss": 3.7854,
+      "step": 3600
+    },
+    {
+      "epoch": 1.2094275536409578,
+      "grad_norm": 0.306640625,
+      "learning_rate": 4.998610735881659e-05,
+      "loss": 3.7765,
+      "step": 3650
+    },
+    {
+      "epoch": 1.2259961892138183,
+      "grad_norm": 0.27734375,
+      "learning_rate": 4.99845769048543e-05,
+      "loss": 3.7835,
+      "step": 3700
+    },
+    {
+      "epoch": 1.2425648247866787,
+      "grad_norm": 0.298828125,
+      "learning_rate": 4.998296652359965e-05,
+      "loss": 3.7809,
+      "step": 3750
+    },
+    {
+      "epoch": 1.2591334603595394,
+      "grad_norm": 0.310546875,
+      "learning_rate": 4.9981276220204344e-05,
+      "loss": 3.7849,
+      "step": 3800
+    },
+    {
+      "epoch": 1.2757020959324,
+      "grad_norm": 0.3125,
+      "learning_rate": 4.997950600007578e-05,
+      "loss": 3.7815,
+      "step": 3850
+    },
+    {
+      "epoch": 1.2922707315052606,
+      "grad_norm": 0.302734375,
+      "learning_rate": 4.997765586887702e-05,
+      "loss": 3.7793,
+      "step": 3900
+    },
+    {
+      "epoch": 1.308839367078121,
+      "grad_norm": 0.30859375,
+      "learning_rate": 4.997572583252672e-05,
+      "loss": 3.7729,
+      "step": 3950
+    },
+    {
+      "epoch": 1.3254080026509816,
+      "grad_norm": 0.29296875,
+      "learning_rate": 4.9973715897199226e-05,
+      "loss": 3.7745,
+      "step": 4000
+    },
+    {
+      "epoch": 1.3419766382238423,
+      "grad_norm": 0.32421875,
+      "learning_rate": 4.9971626069324435e-05,
+      "loss": 3.7688,
+      "step": 4050
+    },
+    {
+      "epoch": 1.358545273796703,
+      "grad_norm": 0.275390625,
+      "learning_rate": 4.996945635558785e-05,
+      "loss": 3.7748,
+      "step": 4100
+    },
+    {
+      "epoch": 1.3751139093695635,
+      "grad_norm": 0.291015625,
+      "learning_rate": 4.996720676293052e-05,
+      "loss": 3.7686,
+      "step": 4150
+    },
+    {
+      "epoch": 1.391682544942424,
+      "grad_norm": 0.287109375,
+      "learning_rate": 4.9964877298549045e-05,
+      "loss": 3.7736,
+      "step": 4200
+    },
+    {
+      "epoch": 1.4082511805152846,
+      "grad_norm": 0.306640625,
+      "learning_rate": 4.9962467969895535e-05,
+      "loss": 3.7751,
+      "step": 4250
+    },
+    {
+      "epoch": 1.4248198160881451,
+      "grad_norm": 0.2890625,
+      "learning_rate": 4.995997878467758e-05,
+      "loss": 3.7673,
+      "step": 4300
+    },
+    {
+      "epoch": 1.4413884516610058,
+      "grad_norm": 0.34375,
+      "learning_rate": 4.995740975085825e-05,
+      "loss": 3.7742,
+      "step": 4350
+    },
+    {
+      "epoch": 1.4579570872338663,
+      "grad_norm": 0.294921875,
+      "learning_rate": 4.9954760876656056e-05,
+      "loss": 3.7737,
+      "step": 4400
+    },
+    {
+      "epoch": 1.4745257228067268,
+      "grad_norm": 0.283203125,
+      "learning_rate": 4.995203217054493e-05,
+      "loss": 3.7704,
+      "step": 4450
+    },
+    {
+      "epoch": 1.4910943583795875,
+      "grad_norm": 0.294921875,
+      "learning_rate": 4.9949223641254156e-05,
+      "loss": 3.7693,
+      "step": 4500
+    },
+    {
+      "epoch": 1.507662993952448,
+      "grad_norm": 0.306640625,
+      "learning_rate": 4.994633529776842e-05,
+      "loss": 3.76,
+      "step": 4550
+    },
+    {
+      "epoch": 1.5242316295253087,
+      "grad_norm": 0.27734375,
+      "learning_rate": 4.994336714932771e-05,
+      "loss": 3.7617,
+      "step": 4600
+    },
+    {
+      "epoch": 1.5408002650981691,
+      "grad_norm": 0.287109375,
+      "learning_rate": 4.9940319205427335e-05,
+      "loss": 3.7737,
+      "step": 4650
+    },
+    {
+      "epoch": 1.5573689006710296,
+      "grad_norm": 0.310546875,
+      "learning_rate": 4.993719147581787e-05,
+      "loss": 3.7699,
+      "step": 4700
+    },
+    {
+      "epoch": 1.5739375362438903,
+      "grad_norm": 0.306640625,
+      "learning_rate": 4.9933983970505116e-05,
+      "loss": 3.7665,
+      "step": 4750
+    },
+    {
+      "epoch": 1.590506171816751,
+      "grad_norm": 0.30859375,
+      "learning_rate": 4.9930696699750095e-05,
+      "loss": 3.7622,
+      "step": 4800
+    },
+    {
+      "epoch": 1.6070748073896115,
+      "grad_norm": 0.287109375,
+      "learning_rate": 4.992732967406901e-05,
+      "loss": 3.7572,
+      "step": 4850
+    },
+    {
+      "epoch": 1.623643442962472,
+      "grad_norm": 0.3125,
+      "learning_rate": 4.992388290423318e-05,
+      "loss": 3.7626,
+      "step": 4900
+    },
+    {
+      "epoch": 1.6402120785353325,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.9920356401269055e-05,
+      "loss": 3.7626,
+      "step": 4950
+    },
+    {
+      "epoch": 1.6567807141081932,
+      "grad_norm": 0.31640625,
+      "learning_rate": 4.991675017645815e-05,
+      "loss": 3.7626,
+      "step": 5000
+    },
+    {
+      "epoch": 1.6567807141081932,
+      "eval_loss": 3.72454833984375,
+      "eval_runtime": 7.9243,
+      "eval_samples_per_second": 122.282,
+      "eval_steps_per_second": 2.019,
+      "step": 5000
+    },
+    {
+      "epoch": 1.6733493496810539,
+      "grad_norm": 0.3046875,
+      "learning_rate": 4.991306424133701e-05,
+      "loss": 3.762,
+      "step": 5050
+    },
+    {
+      "epoch": 1.6899179852539143,
+      "grad_norm": 0.296875,
+      "learning_rate": 4.990929860769719e-05,
+      "loss": 3.7576,
+      "step": 5100
+    },
+    {
+      "epoch": 1.7064866208267748,
+      "grad_norm": 0.3203125,
+      "learning_rate": 4.990545328758518e-05,
+      "loss": 3.7624,
+      "step": 5150
+    },
+    {
+      "epoch": 1.7230552563996355,
+      "grad_norm": 0.302734375,
+      "learning_rate": 4.990152829330243e-05,
+      "loss": 3.757,
+      "step": 5200
+    },
+    {
+      "epoch": 1.7396238919724962,
+      "grad_norm": 0.283203125,
+      "learning_rate": 4.989752363740524e-05,
+      "loss": 3.7655,
+      "step": 5250
+    },
+    {
+      "epoch": 1.7561925275453567,
+      "grad_norm": 0.298828125,
+      "learning_rate": 4.989343933270477e-05,
+      "loss": 3.7575,
+      "step": 5300
+    },
+    {
+      "epoch": 1.7727611631182172,
+      "grad_norm": 0.30859375,
+      "learning_rate": 4.9889275392266984e-05,
+      "loss": 3.7618,
+      "step": 5350
+    },
+    {
+      "epoch": 1.7893297986910777,
+      "grad_norm": 0.30859375,
+      "learning_rate": 4.988503182941259e-05,
+      "loss": 3.7561,
+      "step": 5400
+    },
+    {
+      "epoch": 1.8058984342639384,
+      "grad_norm": 0.298828125,
+      "learning_rate": 4.988070865771702e-05,
+      "loss": 3.7645,
+      "step": 5450
+    },
+    {
+      "epoch": 1.822467069836799,
+      "grad_norm": 0.294921875,
+      "learning_rate": 4.9876305891010385e-05,
+      "loss": 3.7571,
+      "step": 5500
+    },
+    {
+      "epoch": 1.8390357054096595,
+      "grad_norm": 0.30859375,
+      "learning_rate": 4.987182354337744e-05,
+      "loss": 3.7565,
+      "step": 5550
+    },
+    {
+      "epoch": 1.85560434098252,
+      "grad_norm": 0.310546875,
+      "learning_rate": 4.986726162915748e-05,
+      "loss": 3.7604,
+      "step": 5600
+    },
+    {
+      "epoch": 1.8721729765553805,
+      "grad_norm": 0.28125,
+      "learning_rate": 4.9862620162944386e-05,
+      "loss": 3.7497,
+      "step": 5650
+    },
+    {
+      "epoch": 1.8887416121282412,
+      "grad_norm": 0.291015625,
+      "learning_rate": 4.9857899159586496e-05,
+      "loss": 3.7498,
+      "step": 5700
+    },
+    {
+      "epoch": 1.905310247701102,
+      "grad_norm": 0.302734375,
+      "learning_rate": 4.9853098634186625e-05,
+      "loss": 3.7488,
+      "step": 5750
+    },
+    {
+      "epoch": 1.9218788832739624,
+      "grad_norm": 0.29296875,
+      "learning_rate": 4.984821860210196e-05,
+      "loss": 3.7517,
+      "step": 5800
+    },
+    {
+      "epoch": 1.9384475188468229,
+      "grad_norm": 0.3125,
+      "learning_rate": 4.984325907894404e-05,
+      "loss": 3.7454,
+      "step": 5850
+    },
+    {
+      "epoch": 1.9550161544196836,
+      "grad_norm": 0.291015625,
+      "learning_rate": 4.98382200805787e-05,
+      "loss": 3.7446,
+      "step": 5900
+    },
+    {
+      "epoch": 1.9715847899925443,
+      "grad_norm": 0.30078125,
+      "learning_rate": 4.9833101623126034e-05,
+      "loss": 3.751,
+      "step": 5950
+    },
+    {
+      "epoch": 1.9881534255654048,
+      "grad_norm": 0.296875,
+      "learning_rate": 4.982790372296031e-05,
+      "loss": 3.7485,
+      "step": 6000
+    },
+    {
+      "epoch": 2.004639217960401,
+      "grad_norm": 0.341796875,
+      "learning_rate": 4.9822626396709965e-05,
+      "loss": 3.7467,
+      "step": 6050
+    },
+    {
+      "epoch": 2.0212078535332614,
+      "grad_norm": 0.341796875,
+      "learning_rate": 4.98172696612575e-05,
+      "loss": 3.7434,
+      "step": 6100
+    },
+    {
+      "epoch": 2.0377764891061223,
+      "grad_norm": 0.330078125,
+      "learning_rate": 4.981183353373946e-05,
+      "loss": 3.7451,
+      "step": 6150
+    },
+    {
+      "epoch": 2.054345124678983,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.980631803154638e-05,
+      "loss": 3.7338,
+      "step": 6200
+    },
+    {
+      "epoch": 2.0709137602518433,
+      "grad_norm": 0.333984375,
+      "learning_rate": 4.98007231723227e-05,
+      "loss": 3.7348,
+      "step": 6250
+    },
+    {
+      "epoch": 2.0874823958247037,
+      "grad_norm": 0.3359375,
+      "learning_rate": 4.979504897396675e-05,
+      "loss": 3.7295,
+      "step": 6300
+    },
+    {
+      "epoch": 2.104051031397564,
+      "grad_norm": 0.365234375,
+      "learning_rate": 4.978929545463066e-05,
+      "loss": 3.7357,
+      "step": 6350
+    },
+    {
+      "epoch": 2.120619666970425,
+      "grad_norm": 0.365234375,
+      "learning_rate": 4.97834626327203e-05,
+      "loss": 3.7387,
+      "step": 6400
+    },
+    {
+      "epoch": 2.1371883025432856,
+      "grad_norm": 0.345703125,
+      "learning_rate": 4.9777550526895265e-05,
+      "loss": 3.7359,
+      "step": 6450
+    },
+    {
+      "epoch": 2.153756938116146,
+      "grad_norm": 0.357421875,
+      "learning_rate": 4.977155915606877e-05,
+      "loss": 3.7359,
+      "step": 6500
+    },
+    {
+      "epoch": 2.1703255736890066,
+      "grad_norm": 0.330078125,
+      "learning_rate": 4.9765488539407586e-05,
+      "loss": 3.7333,
+      "step": 6550
+    },
+    {
+      "epoch": 2.186894209261867,
+      "grad_norm": 0.33203125,
+      "learning_rate": 4.975933869633202e-05,
+      "loss": 3.7296,
+      "step": 6600
+    },
+    {
+      "epoch": 2.203462844834728,
+      "grad_norm": 0.3125,
+      "learning_rate": 4.9753109646515814e-05,
+      "loss": 3.7297,
+      "step": 6650
+    },
+    {
+      "epoch": 2.2200314804075885,
+      "grad_norm": 0.361328125,
+      "learning_rate": 4.974680140988612e-05,
+      "loss": 3.7334,
+      "step": 6700
+    },
+    {
+      "epoch": 2.236600115980449,
+      "grad_norm": 0.33984375,
+      "learning_rate": 4.974041400662338e-05,
+      "loss": 3.7342,
+      "step": 6750
+    },
+    {
+      "epoch": 2.2531687515533094,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.973394745716133e-05,
+      "loss": 3.726,
+      "step": 6800
+    },
+    {
+      "epoch": 2.2697373871261703,
+      "grad_norm": 0.35546875,
+      "learning_rate": 4.972740178218688e-05,
+      "loss": 3.7359,
+      "step": 6850
+    },
+    {
+      "epoch": 2.286306022699031,
+      "grad_norm": 0.322265625,
+      "learning_rate": 4.972077700264007e-05,
+      "loss": 3.7298,
+      "step": 6900
+    },
+    {
+      "epoch": 2.3028746582718913,
+      "grad_norm": 0.326171875,
+      "learning_rate": 4.9714073139714004e-05,
+      "loss": 3.731,
+      "step": 6950
+    },
+    {
+      "epoch": 2.319443293844752,
+      "grad_norm": 0.337890625,
+      "learning_rate": 4.970729021485476e-05,
+      "loss": 3.7268,
+      "step": 7000
+    },
+    {
+      "epoch": 2.3360119294176123,
+      "grad_norm": 0.33203125,
+      "learning_rate": 4.9700428249761386e-05,
+      "loss": 3.7249,
+      "step": 7050
+    },
+    {
+      "epoch": 2.352580564990473,
+      "grad_norm": 0.33203125,
+      "learning_rate": 4.969348726638574e-05,
+      "loss": 3.7297,
+      "step": 7100
+    },
+    {
+      "epoch": 2.3691492005633337,
+      "grad_norm": 0.34765625,
+      "learning_rate": 4.968646728693248e-05,
+      "loss": 3.728,
+      "step": 7150
+    },
+    {
+      "epoch": 2.385717836136194,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.967936833385898e-05,
+      "loss": 3.7298,
+      "step": 7200
+    },
+    {
+      "epoch": 2.4022864717090546,
+      "grad_norm": 0.357421875,
+      "learning_rate": 4.9672190429875266e-05,
+      "loss": 3.7228,
+      "step": 7250
+    },
+    {
+      "epoch": 2.4188551072819156,
+      "grad_norm": 0.3671875,
+      "learning_rate": 4.96649335979439e-05,
+      "loss": 3.7254,
+      "step": 7300
+    },
+    {
+      "epoch": 2.435423742854776,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.9657597861279976e-05,
+      "loss": 3.7284,
+      "step": 7350
+    },
+    {
+      "epoch": 2.4519923784276365,
+      "grad_norm": 0.31640625,
+      "learning_rate": 4.965018324335099e-05,
+      "loss": 3.7213,
+      "step": 7400
+    },
+    {
+      "epoch": 2.468561014000497,
+      "grad_norm": 0.310546875,
+      "learning_rate": 4.964268976787679e-05,
+      "loss": 3.7224,
+      "step": 7450
+    },
+    {
+      "epoch": 2.4851296495733575,
+      "grad_norm": 0.330078125,
+      "learning_rate": 4.9635117458829496e-05,
+      "loss": 3.7222,
+      "step": 7500
+    },
+    {
+      "epoch": 2.501698285146218,
+      "grad_norm": 0.322265625,
+      "learning_rate": 4.962746634043341e-05,
+      "loss": 3.7286,
+      "step": 7550
+    },
+    {
+      "epoch": 2.518266920719079,
+      "grad_norm": 0.37890625,
+      "learning_rate": 4.961973643716497e-05,
+      "loss": 3.7287,
+      "step": 7600
+    },
+    {
+      "epoch": 2.5348355562919394,
+      "grad_norm": 0.326171875,
+      "learning_rate": 4.961192777375263e-05,
+      "loss": 3.7208,
+      "step": 7650
+    },
+    {
+      "epoch": 2.5514041918648,
+      "grad_norm": 0.3515625,
+      "learning_rate": 4.9604040375176816e-05,
+      "loss": 3.7305,
+      "step": 7700
+    },
+    {
+      "epoch": 2.5679728274376608,
+      "grad_norm": 0.330078125,
+      "learning_rate": 4.9596074266669844e-05,
+      "loss": 3.7171,
+      "step": 7750
+    },
+    {
+      "epoch": 2.5845414630105212,
+      "grad_norm": 0.32421875,
+      "learning_rate": 4.95880294737158e-05,
+      "loss": 3.721,
+      "step": 7800
+    },
+    {
+      "epoch": 2.6011100985833817,
+      "grad_norm": 0.310546875,
+      "learning_rate": 4.9579906022050517e-05,
+      "loss": 3.7225,
+      "step": 7850
+    },
+    {
+      "epoch": 2.617678734156242,
+      "grad_norm": 0.322265625,
+      "learning_rate": 4.957170393766143e-05,
+      "loss": 3.7215,
+      "step": 7900
+    },
+    {
+      "epoch": 2.6342473697291027,
+      "grad_norm": 0.345703125,
+      "learning_rate": 4.956342324678755e-05,
+      "loss": 3.7237,
+      "step": 7950
+    },
+    {
+      "epoch": 2.650816005301963,
+      "grad_norm": 0.3125,
+      "learning_rate": 4.9555063975919345e-05,
+      "loss": 3.7189,
+      "step": 8000
+    },
+    {
+      "epoch": 2.667384640874824,
+      "grad_norm": 0.314453125,
+      "learning_rate": 4.954662615179868e-05,
+      "loss": 3.7209,
+      "step": 8050
+    },
+    {
+      "epoch": 2.6839532764476846,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.953810980141869e-05,
+      "loss": 3.7267,
+      "step": 8100
+    },
+    {
+      "epoch": 2.700521912020545,
+      "grad_norm": 0.32421875,
+      "learning_rate": 4.952951495202374e-05,
+      "loss": 3.72,
+      "step": 8150
+    },
+    {
+      "epoch": 2.717090547593406,
+      "grad_norm": 0.33203125,
+      "learning_rate": 4.9520841631109315e-05,
+      "loss": 3.7168,
+      "step": 8200
+    },
+    {
+      "epoch": 2.7336591831662664,
+      "grad_norm": 0.33203125,
+      "learning_rate": 4.951208986642194e-05,
+      "loss": 3.7167,
+      "step": 8250
+    },
+    {
+      "epoch": 2.750227818739127,
+      "grad_norm": 0.32421875,
+      "learning_rate": 4.9503259685959074e-05,
+      "loss": 3.7192,
+      "step": 8300
+    },
+    {
+      "epoch": 2.7667964543119874,
+      "grad_norm": 0.3515625,
+      "learning_rate": 4.949435111796905e-05,
+      "loss": 3.7162,
+      "step": 8350
+    },
+    {
+      "epoch": 2.783365089884848,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.948536419095095e-05,
+      "loss": 3.7161,
+      "step": 8400
+    },
+    {
+      "epoch": 2.7999337254577084,
+      "grad_norm": 0.32421875,
+      "learning_rate": 4.947629893365453e-05,
+      "loss": 3.718,
+      "step": 8450
+    },
+    {
+      "epoch": 2.8165023610305693,
+      "grad_norm": 0.322265625,
+      "learning_rate": 4.9467155375080165e-05,
+      "loss": 3.7152,
+      "step": 8500
+    },
+    {
+      "epoch": 2.8330709966034298,
+      "grad_norm": 0.34375,
+      "learning_rate": 4.9457933544478684e-05,
+      "loss": 3.7237,
+      "step": 8550
+    },
+    {
+      "epoch": 2.8496396321762902,
+      "grad_norm": 0.330078125,
+      "learning_rate": 4.944863347135132e-05,
+      "loss": 3.7092,
+      "step": 8600
+    },
+    {
+      "epoch": 2.8662082677491507,
+      "grad_norm": 0.328125,
+      "learning_rate": 4.943925518544962e-05,
+      "loss": 3.7174,
+      "step": 8650
+    },
+    {
+      "epoch": 2.8827769033220116,
+      "grad_norm": 0.3359375,
+      "learning_rate": 4.942979871677532e-05,
+      "loss": 3.7162,
+      "step": 8700
+    },
+    {
+      "epoch": 2.899345538894872,
+      "grad_norm": 0.359375,
+      "learning_rate": 4.94202640955803e-05,
+      "loss": 3.7152,
+      "step": 8750
+    },
+    {
+      "epoch": 2.9159141744677326,
+      "grad_norm": 0.357421875,
+      "learning_rate": 4.9410651352366435e-05,
+      "loss": 3.7181,
+      "step": 8800
+    },
+    {
+      "epoch": 2.932482810040593,
+      "grad_norm": 0.333984375,
+      "learning_rate": 4.94009605178855e-05,
+      "loss": 3.7177,
+      "step": 8850
+    },
+    {
+      "epoch": 2.9490514456134536,
+      "grad_norm": 0.341796875,
+      "learning_rate": 4.939119162313912e-05,
+      "loss": 3.7216,
+      "step": 8900
+    },
+    {
+      "epoch": 2.965620081186314,
+      "grad_norm": 0.3671875,
+      "learning_rate": 4.9381344699378626e-05,
+      "loss": 3.712,
+      "step": 8950
+    },
+    {
+      "epoch": 2.982188716759175,
+      "grad_norm": 0.322265625,
+      "learning_rate": 4.937141977810497e-05,
+      "loss": 3.7199,
+      "step": 9000
+    },
+    {
+      "epoch": 2.9987573523320354,
+      "grad_norm": 0.326171875,
+      "learning_rate": 4.936141689106861e-05,
+      "loss": 3.7058,
+      "step": 9050
+    },
+    {
+      "epoch": 3.0152431447270316,
+      "grad_norm": 0.408203125,
+      "learning_rate": 4.935133607026945e-05,
+      "loss": 3.6969,
+      "step": 9100
+    },
+    {
+      "epoch": 3.0318117802998925,
+      "grad_norm": 0.41796875,
+      "learning_rate": 4.934117734795669e-05,
+      "loss": 3.706,
+      "step": 9150
+    },
+    {
+      "epoch": 3.048380415872753,
+      "grad_norm": 0.369140625,
+      "learning_rate": 4.933094075662874e-05,
+      "loss": 3.7006,
+      "step": 9200
+    },
+    {
+      "epoch": 3.0649490514456135,
+      "grad_norm": 0.361328125,
+      "learning_rate": 4.9320626329033134e-05,
+      "loss": 3.703,
+      "step": 9250
+    },
+    {
+      "epoch": 3.081517687018474,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.9310234098166396e-05,
+      "loss": 3.7084,
+      "step": 9300
+    },
+    {
+      "epoch": 3.0980863225913344,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.929976409727395e-05,
+      "loss": 3.6991,
+      "step": 9350
+    },
+    {
+      "epoch": 3.1146549581641954,
+      "grad_norm": 0.40234375,
+      "learning_rate": 4.928921635985001e-05,
+      "loss": 3.7057,
+      "step": 9400
+    },
+    {
+      "epoch": 3.131223593737056,
+      "grad_norm": 0.392578125,
+      "learning_rate": 4.9278590919637466e-05,
+      "loss": 3.703,
+      "step": 9450
+    },
+    {
+      "epoch": 3.1477922293099163,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.9267887810627824e-05,
+      "loss": 3.7103,
+      "step": 9500
+    },
+    {
+      "epoch": 3.164360864882777,
+      "grad_norm": 0.359375,
+      "learning_rate": 4.9257107067061e-05,
+      "loss": 3.7017,
+      "step": 9550
+    },
+    {
+      "epoch": 3.1809295004556373,
+      "grad_norm": 0.3828125,
+      "learning_rate": 4.924624872342531e-05,
+      "loss": 3.6937,
+      "step": 9600
+    },
+    {
+      "epoch": 3.197498136028498,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.92353128144573e-05,
+      "loss": 3.7063,
+      "step": 9650
+    },
+    {
+      "epoch": 3.2140667716013587,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.9224299375141656e-05,
+      "loss": 3.6978,
+      "step": 9700
+    },
+    {
+      "epoch": 3.230635407174219,
+      "grad_norm": 0.35546875,
+      "learning_rate": 4.921320844071109e-05,
+      "loss": 3.705,
+      "step": 9750
+    },
+    {
+      "epoch": 3.2472040427470796,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.920204004664624e-05,
+      "loss": 3.704,
+      "step": 9800
+    },
+    {
+      "epoch": 3.26377267831994,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.91907942286755e-05,
+      "loss": 3.703,
+      "step": 9850
+    },
+    {
+      "epoch": 3.280341313892801,
+      "grad_norm": 0.37109375,
+      "learning_rate": 4.917947102277499e-05,
+      "loss": 3.7027,
+      "step": 9900
+    },
+    {
+      "epoch": 3.2969099494656615,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.916807046516838e-05,
+      "loss": 3.6977,
+      "step": 9950
+    },
+    {
+      "epoch": 3.313478585038522,
+      "grad_norm": 0.359375,
+      "learning_rate": 4.9156592592326814e-05,
+      "loss": 3.6968,
+      "step": 10000
+    },
+    {
+      "epoch": 3.313478585038522,
+      "eval_loss": 3.6868932247161865,
+      "eval_runtime": 8.062,
+      "eval_samples_per_second": 120.194,
+      "eval_steps_per_second": 1.985,
+      "step": 10000
+    },
+    {
+      "epoch": 3.3300472206113825,
+      "grad_norm": 0.37890625,
+      "learning_rate": 4.9145037440968746e-05,
+      "loss": 3.7025,
+      "step": 10050
+    },
+    {
+      "epoch": 3.3466158561842434,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.913340504805984e-05,
+      "loss": 3.7034,
+      "step": 10100
+    },
+    {
+      "epoch": 3.363184491757104,
+      "grad_norm": 0.357421875,
+      "learning_rate": 4.912169545081292e-05,
+      "loss": 3.7025,
+      "step": 10150
+    },
+    {
+      "epoch": 3.3797531273299644,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.910990868668772e-05,
+      "loss": 3.6956,
+      "step": 10200
+    },
+    {
+      "epoch": 3.396321762902825,
+      "grad_norm": 0.3828125,
+      "learning_rate": 4.9098044793390876e-05,
+      "loss": 3.7025,
+      "step": 10250
+    },
+    {
+      "epoch": 3.4128903984756853,
+      "grad_norm": 0.36328125,
+      "learning_rate": 4.908610380887576e-05,
+      "loss": 3.6915,
+      "step": 10300
+    },
+    {
+      "epoch": 3.4294590340485462,
+      "grad_norm": 0.384765625,
+      "learning_rate": 4.9074085771342365e-05,
+      "loss": 3.6987,
+      "step": 10350
+    },
+    {
+      "epoch": 3.4460276696214067,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.9061990719237175e-05,
+      "loss": 3.6943,
+      "step": 10400
+    },
+    {
+      "epoch": 3.462596305194267,
+      "grad_norm": 0.380859375,
+      "learning_rate": 4.904981869125303e-05,
+      "loss": 3.6961,
+      "step": 10450
+    },
+    {
+      "epoch": 3.4791649407671277,
+      "grad_norm": 0.408203125,
+      "learning_rate": 4.903756972632908e-05,
+      "loss": 3.6985,
+      "step": 10500
+    },
+    {
+      "epoch": 3.4957335763399886,
+      "grad_norm": 0.3828125,
+      "learning_rate": 4.9025243863650545e-05,
+      "loss": 3.7024,
+      "step": 10550
+    },
+    {
+      "epoch": 3.512302211912849,
+      "grad_norm": 0.365234375,
+      "learning_rate": 4.901284114264866e-05,
+      "loss": 3.6989,
+      "step": 10600
+    },
+    {
+      "epoch": 3.5288708474857096,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.900036160300053e-05,
+      "loss": 3.7015,
+      "step": 10650
+    },
+    {
+      "epoch": 3.54543948305857,
+      "grad_norm": 0.365234375,
+      "learning_rate": 4.898780528462904e-05,
+      "loss": 3.6958,
+      "step": 10700
+    },
+    {
+      "epoch": 3.5620081186314305,
+      "grad_norm": 0.392578125,
+      "learning_rate": 4.8975172227702665e-05,
+      "loss": 3.6988,
+      "step": 10750
+    },
+    {
+      "epoch": 3.578576754204291,
+      "grad_norm": 0.365234375,
+      "learning_rate": 4.896246247263537e-05,
+      "loss": 3.6926,
+      "step": 10800
+    },
+    {
+      "epoch": 3.595145389777152,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.89496760600865e-05,
+      "loss": 3.6903,
+      "step": 10850
+    },
+    {
+      "epoch": 3.6117140253500124,
+      "grad_norm": 0.384765625,
+      "learning_rate": 4.893681303096062e-05,
+      "loss": 3.6945,
+      "step": 10900
+    },
+    {
+      "epoch": 3.628282660922873,
+      "grad_norm": 0.369140625,
+      "learning_rate": 4.892387342640739e-05,
+      "loss": 3.6977,
+      "step": 10950
+    },
+    {
+      "epoch": 3.644851296495734,
+      "grad_norm": 0.365234375,
+      "learning_rate": 4.8910857287821465e-05,
+      "loss": 3.6882,
+      "step": 11000
+    },
+    {
+      "epoch": 3.6614199320685943,
+      "grad_norm": 0.37109375,
+      "learning_rate": 4.889776465684232e-05,
+      "loss": 3.6968,
+      "step": 11050
+    },
+    {
+      "epoch": 3.6779885676414548,
+      "grad_norm": 0.359375,
+      "learning_rate": 4.888459557535413e-05,
+      "loss": 3.6937,
+      "step": 11100
+    },
+    {
+      "epoch": 3.6945572032143152,
+      "grad_norm": 0.375,
+      "learning_rate": 4.8871350085485646e-05,
+      "loss": 3.7016,
+      "step": 11150
+    },
+    {
+      "epoch": 3.7111258387871757,
+      "grad_norm": 0.384765625,
+      "learning_rate": 4.8858028229610055e-05,
+      "loss": 3.6983,
+      "step": 11200
+    },
+    {
+      "epoch": 3.727694474360036,
+      "grad_norm": 0.375,
+      "learning_rate": 4.884463005034485e-05,
+      "loss": 3.6995,
+      "step": 11250
+    },
+    {
+      "epoch": 3.744263109932897,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.8831155590551674e-05,
+      "loss": 3.6978,
+      "step": 11300
+    },
+    {
+      "epoch": 3.7608317455057576,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.881760489333621e-05,
+      "loss": 3.6981,
+      "step": 11350
+    },
+    {
+      "epoch": 3.777400381078618,
+      "grad_norm": 0.375,
+      "learning_rate": 4.880397800204801e-05,
+      "loss": 3.6884,
+      "step": 11400
+    },
+    {
+      "epoch": 3.793969016651479,
+      "grad_norm": 0.375,
+      "learning_rate": 4.879027496028041e-05,
+      "loss": 3.6957,
+      "step": 11450
+    },
+    {
+      "epoch": 3.8105376522243395,
+      "grad_norm": 0.384765625,
+      "learning_rate": 4.8776495811870315e-05,
+      "loss": 3.6933,
+      "step": 11500
+    },
+    {
+      "epoch": 3.8271062877972,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.876264060089813e-05,
+      "loss": 3.6857,
+      "step": 11550
+    },
+    {
+      "epoch": 3.8436749233700604,
+      "grad_norm": 0.357421875,
+      "learning_rate": 4.874870937168757e-05,
+      "loss": 3.6998,
+      "step": 11600
+    },
+    {
+      "epoch": 3.860243558942921,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.873470216880557e-05,
+      "loss": 3.6876,
+      "step": 11650
+    },
+    {
+      "epoch": 3.8768121945157814,
+      "grad_norm": 0.375,
+      "learning_rate": 4.8720619037062056e-05,
+      "loss": 3.6933,
+      "step": 11700
+    },
+    {
+      "epoch": 3.8933808300886423,
+      "grad_norm": 0.380859375,
+      "learning_rate": 4.87064600215099e-05,
+      "loss": 3.6993,
+      "step": 11750
+    },
+    {
+      "epoch": 3.909949465661503,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.869222516744473e-05,
+      "loss": 3.6934,
+      "step": 11800
+    },
+    {
+      "epoch": 3.9265181012343633,
+      "grad_norm": 0.392578125,
+      "learning_rate": 4.867791452040475e-05,
+      "loss": 3.6912,
+      "step": 11850
+    },
+    {
+      "epoch": 3.9430867368072238,
+      "grad_norm": 0.359375,
+      "learning_rate": 4.866352812617066e-05,
+      "loss": 3.6898,
+      "step": 11900
+    },
+    {
+      "epoch": 3.9596553723800847,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.864906603076549e-05,
+      "loss": 3.6901,
+      "step": 11950
+    },
+    {
+      "epoch": 3.976224007952945,
+      "grad_norm": 0.3671875,
+      "learning_rate": 4.8634528280454404e-05,
+      "loss": 3.6925,
+      "step": 12000
+    },
+    {
+      "epoch": 3.9927926435258057,
+      "grad_norm": 0.365234375,
+      "learning_rate": 4.861991492174464e-05,
+      "loss": 3.6884,
+      "step": 12050
+    },
+    {
+      "epoch": 4.009278435920802,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.8605226001385256e-05,
+      "loss": 3.685,
+      "step": 12100
+    },
+    {
+      "epoch": 4.025847071493662,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.85904615663671e-05,
+      "loss": 3.6815,
+      "step": 12150
+    },
+    {
+      "epoch": 4.042415707066523,
+      "grad_norm": 0.43359375,
+      "learning_rate": 4.857562166392253e-05,
+      "loss": 3.6793,
+      "step": 12200
+    },
+    {
+      "epoch": 4.058984342639383,
+      "grad_norm": 0.443359375,
+      "learning_rate": 4.8560706341525386e-05,
+      "loss": 3.6799,
+      "step": 12250
+    },
+    {
+      "epoch": 4.075552978212245,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.854571564689075e-05,
+      "loss": 3.6809,
+      "step": 12300
+    },
+    {
+      "epoch": 4.092121613785105,
+      "grad_norm": 0.431640625,
+      "learning_rate": 4.853064962797483e-05,
+      "loss": 3.6788,
+      "step": 12350
+    },
+    {
+      "epoch": 4.108690249357966,
+      "grad_norm": 0.419921875,
+      "learning_rate": 4.851550833297481e-05,
+      "loss": 3.6869,
+      "step": 12400
+    },
+    {
+      "epoch": 4.125258884930826,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.850029181032869e-05,
+      "loss": 3.6864,
+      "step": 12450
+    },
+    {
+      "epoch": 4.1418275205036865,
+      "grad_norm": 0.4453125,
+      "learning_rate": 4.8485000108715094e-05,
+      "loss": 3.6845,
+      "step": 12500
+    },
+    {
+      "epoch": 4.158396156076547,
+      "grad_norm": 0.4375,
+      "learning_rate": 4.846963327705319e-05,
+      "loss": 3.6772,
+      "step": 12550
+    },
+    {
+      "epoch": 4.1749647916494075,
+      "grad_norm": 0.462890625,
+      "learning_rate": 4.845419136450248e-05,
+      "loss": 3.691,
+      "step": 12600
+    },
+    {
+      "epoch": 4.191533427222268,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.843867442046264e-05,
+      "loss": 3.6889,
+      "step": 12650
+    },
+    {
+      "epoch": 4.208102062795128,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.842308249457339e-05,
+      "loss": 3.6835,
+      "step": 12700
+    },
+    {
+      "epoch": 4.22467069836799,
+      "grad_norm": 0.4296875,
+      "learning_rate": 4.8407415636714315e-05,
+      "loss": 3.6833,
+      "step": 12750
+    },
+    {
+      "epoch": 4.24123933394085,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.839167389700472e-05,
+      "loss": 3.6782,
+      "step": 12800
+    },
+    {
+      "epoch": 4.257807969513711,
+      "grad_norm": 0.41796875,
+      "learning_rate": 4.837585732580345e-05,
+      "loss": 3.6887,
+      "step": 12850
+    },
+    {
+      "epoch": 4.274376605086571,
+      "grad_norm": 0.416015625,
+      "learning_rate": 4.835996597370877e-05,
+      "loss": 3.6855,
+      "step": 12900
+    },
+    {
+      "epoch": 4.290945240659432,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.8343999891558136e-05,
+      "loss": 3.6828,
+      "step": 12950
+    },
+    {
+      "epoch": 4.307513876232292,
+      "grad_norm": 0.44140625,
+      "learning_rate": 4.832795913042809e-05,
+      "loss": 3.6815,
+      "step": 13000
+    },
+    {
+      "epoch": 4.324082511805153,
+      "grad_norm": 0.4296875,
+      "learning_rate": 4.831184374163407e-05,
+      "loss": 3.6836,
+      "step": 13050
+    },
+    {
+      "epoch": 4.340651147378013,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.829565377673026e-05,
+      "loss": 3.6919,
+      "step": 13100
+    },
+    {
+      "epoch": 4.357219782950874,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.827938928750941e-05,
+      "loss": 3.6815,
+      "step": 13150
+    },
+    {
+      "epoch": 4.373788418523734,
+      "grad_norm": 0.431640625,
+      "learning_rate": 4.82630503260027e-05,
+      "loss": 3.6834,
+      "step": 13200
+    },
+    {
+      "epoch": 4.3903570540965955,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.824663694447952e-05,
+      "loss": 3.6834,
+      "step": 13250
+    },
+    {
+      "epoch": 4.406925689669456,
+      "grad_norm": 0.4296875,
+      "learning_rate": 4.823014919544734e-05,
+      "loss": 3.6814,
+      "step": 13300
+    },
+    {
+      "epoch": 4.4234943252423164,
+      "grad_norm": 0.431640625,
+      "learning_rate": 4.8213587131651566e-05,
+      "loss": 3.6849,
+      "step": 13350
+    },
+    {
+      "epoch": 4.440062960815177,
+      "grad_norm": 0.4453125,
+      "learning_rate": 4.819695080607531e-05,
+      "loss": 3.6795,
+      "step": 13400
+    },
+    {
+      "epoch": 4.456631596388037,
+      "grad_norm": 0.4375,
+      "learning_rate": 4.8180240271939256e-05,
+      "loss": 3.6823,
+      "step": 13450
+    },
+    {
+      "epoch": 4.473200231960898,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.8163455582701514e-05,
+      "loss": 3.6798,
+      "step": 13500
+    },
+    {
+      "epoch": 4.489768867533758,
+      "grad_norm": 0.416015625,
+      "learning_rate": 4.8146596792057385e-05,
+      "loss": 3.6811,
+      "step": 13550
+    },
+    {
+      "epoch": 4.506337503106619,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.812966395393924e-05,
+      "loss": 3.6858,
+      "step": 13600
+    },
+    {
+      "epoch": 4.522906138679479,
+      "grad_norm": 0.44140625,
+      "learning_rate": 4.811265712251635e-05,
+      "loss": 3.6776,
+      "step": 13650
+    },
+    {
+      "epoch": 4.539474774252341,
+      "grad_norm": 0.40625,
+      "learning_rate": 4.809557635219465e-05,
+      "loss": 3.6812,
+      "step": 13700
+    },
+    {
+      "epoch": 4.556043409825201,
+      "grad_norm": 0.416015625,
+      "learning_rate": 4.807842169761667e-05,
+      "loss": 3.6772,
+      "step": 13750
+    },
+    {
+      "epoch": 4.572612045398062,
+      "grad_norm": 0.43359375,
+      "learning_rate": 4.806119321366124e-05,
+      "loss": 3.6846,
+      "step": 13800
+    },
+    {
+      "epoch": 4.589180680970922,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.8043890955443436e-05,
+      "loss": 3.6826,
+      "step": 13850
+    },
+    {
+      "epoch": 4.605749316543783,
+      "grad_norm": 0.447265625,
+      "learning_rate": 4.802651497831428e-05,
+      "loss": 3.6789,
+      "step": 13900
+    },
+    {
+      "epoch": 4.622317952116643,
+      "grad_norm": 0.41015625,
+      "learning_rate": 4.8009065337860685e-05,
+      "loss": 3.6776,
+      "step": 13950
+    },
+    {
+      "epoch": 4.638886587689504,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.799154208990518e-05,
+      "loss": 3.6736,
+      "step": 14000
+    },
+    {
+      "epoch": 4.655455223262364,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.7973945290505766e-05,
+      "loss": 3.6817,
+      "step": 14050
+    },
+    {
+      "epoch": 4.6720238588352245,
+      "grad_norm": 0.46484375,
+      "learning_rate": 4.7956274995955775e-05,
+      "loss": 3.6843,
+      "step": 14100
+    },
+    {
+      "epoch": 4.688592494408086,
+      "grad_norm": 0.443359375,
+      "learning_rate": 4.793853126278361e-05,
+      "loss": 3.6777,
+      "step": 14150
+    },
+    {
+      "epoch": 4.705161129980946,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.792071414775265e-05,
+      "loss": 3.6788,
+      "step": 14200
+    },
+    {
+      "epoch": 4.721729765553807,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.7902823707861e-05,
+      "loss": 3.6818,
+      "step": 14250
+    },
+    {
+      "epoch": 4.738298401126667,
+      "grad_norm": 0.451171875,
+      "learning_rate": 4.7884860000341344e-05,
+      "loss": 3.6844,
+      "step": 14300
+    },
+    {
+      "epoch": 4.754867036699528,
+      "grad_norm": 0.4375,
+      "learning_rate": 4.786682308266076e-05,
+      "loss": 3.68,
+      "step": 14350
+    },
+    {
+      "epoch": 4.771435672272388,
+      "grad_norm": 0.4296875,
+      "learning_rate": 4.784871301252052e-05,
+      "loss": 3.6782,
+      "step": 14400
+    },
+    {
+      "epoch": 4.788004307845249,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.783052984785593e-05,
+      "loss": 3.6743,
+      "step": 14450
+    },
+    {
+      "epoch": 4.804572943418109,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.781227364683611e-05,
+      "loss": 3.6864,
+      "step": 14500
+    },
+    {
+      "epoch": 4.82114157899097,
+      "grad_norm": 0.423828125,
+      "learning_rate": 4.7793944467863836e-05,
+      "loss": 3.6766,
+      "step": 14550
+    },
+    {
+      "epoch": 4.837710214563831,
+      "grad_norm": 0.41796875,
+      "learning_rate": 4.777554236957537e-05,
+      "loss": 3.676,
+      "step": 14600
+    },
+    {
+      "epoch": 4.854278850136692,
+      "grad_norm": 0.408203125,
+      "learning_rate": 4.775706741084019e-05,
+      "loss": 3.6834,
+      "step": 14650
+    },
+    {
+      "epoch": 4.870847485709552,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.7738519650760935e-05,
+      "loss": 3.679,
+      "step": 14700
+    },
+    {
+      "epoch": 4.8874161212824125,
+      "grad_norm": 0.41796875,
+      "learning_rate": 4.7719899148673074e-05,
+      "loss": 3.6792,
+      "step": 14750
+    },
+    {
+      "epoch": 4.903984756855273,
+      "grad_norm": 0.431640625,
+      "learning_rate": 4.770120596414481e-05,
+      "loss": 3.6783,
+      "step": 14800
+    },
+    {
+      "epoch": 4.9205533924281335,
+      "grad_norm": 0.44921875,
+      "learning_rate": 4.768244015697687e-05,
+      "loss": 3.6849,
+      "step": 14850
+    },
+    {
+      "epoch": 4.937122028000994,
+      "grad_norm": 0.462890625,
+      "learning_rate": 4.7663601787202285e-05,
+      "loss": 3.6749,
+      "step": 14900
+    },
+    {
+      "epoch": 4.9536906635738545,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.7644690915086245e-05,
+      "loss": 3.6825,
+      "step": 14950
+    },
+    {
+      "epoch": 4.970259299146715,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.7625707601125845e-05,
+      "loss": 3.6747,
+      "step": 15000
+    },
+    {
+      "epoch": 4.970259299146715,
+      "eval_loss": 3.661933422088623,
+      "eval_runtime": 7.9724,
+      "eval_samples_per_second": 121.545,
+      "eval_steps_per_second": 2.007,
+      "step": 15000
+    },
+    {
+      "epoch": 4.986827934719576,
+      "grad_norm": 0.408203125,
+      "learning_rate": 4.760665190604996e-05,
+      "loss": 3.6823,
+      "step": 15050
+    },
+    {
+      "epoch": 5.0033137271145725,
+      "grad_norm": 0.47265625,
+      "learning_rate": 4.758752389081901e-05,
+      "loss": 3.6759,
+      "step": 15100
+    },
+    {
+      "epoch": 5.019882362687433,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.756832361662473e-05,
+      "loss": 3.6767,
+      "step": 15150
+    },
+    {
+      "epoch": 5.036450998260293,
+      "grad_norm": 0.486328125,
+      "learning_rate": 4.754905114489009e-05,
+      "loss": 3.6778,
+      "step": 15200
+    },
+    {
+      "epoch": 5.053019633833154,
+      "grad_norm": 0.470703125,
+      "learning_rate": 4.752970653726896e-05,
+      "loss": 3.6676,
+      "step": 15250
+    },
+    {
+      "epoch": 5.069588269406014,
+      "grad_norm": 0.478515625,
+      "learning_rate": 4.751028985564602e-05,
+      "loss": 3.6659,
+      "step": 15300
+    },
+    {
+      "epoch": 5.086156904978875,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.7490801162136484e-05,
+      "loss": 3.6648,
+      "step": 15350
+    },
+    {
+      "epoch": 5.102725540551735,
+      "grad_norm": 0.478515625,
+      "learning_rate": 4.7471240519085964e-05,
+      "loss": 3.6797,
+      "step": 15400
+    },
+    {
+      "epoch": 5.119294176124596,
+      "grad_norm": 0.4765625,
+      "learning_rate": 4.745160798907023e-05,
+      "loss": 3.6685,
+      "step": 15450
+    },
+    {
+      "epoch": 5.135862811697456,
+      "grad_norm": 0.48046875,
+      "learning_rate": 4.7431903634895034e-05,
+      "loss": 3.6657,
+      "step": 15500
+    },
+    {
+      "epoch": 5.152431447270318,
+      "grad_norm": 0.484375,
+      "learning_rate": 4.741212751959589e-05,
+      "loss": 3.6699,
+      "step": 15550
+    },
+    {
+      "epoch": 5.169000082843178,
+      "grad_norm": 0.466796875,
+      "learning_rate": 4.739227970643787e-05,
+      "loss": 3.6705,
+      "step": 15600
+    },
+    {
+      "epoch": 5.185568718416039,
+      "grad_norm": 0.4921875,
+      "learning_rate": 4.737236025891544e-05,
+      "loss": 3.6692,
+      "step": 15650
+    },
+    {
+      "epoch": 5.202137353988899,
+      "grad_norm": 0.4921875,
+      "learning_rate": 4.73523692407522e-05,
+      "loss": 3.6632,
+      "step": 15700
+    },
+    {
+      "epoch": 5.21870598956176,
+      "grad_norm": 0.470703125,
+      "learning_rate": 4.733230671590074e-05,
+      "loss": 3.6764,
+      "step": 15750
+    },
+    {
+      "epoch": 5.23527462513462,
+      "grad_norm": 0.4921875,
+      "learning_rate": 4.731217274854236e-05,
+      "loss": 3.6778,
+      "step": 15800
+    },
+    {
+      "epoch": 5.2518432607074805,
+      "grad_norm": 0.482421875,
+      "learning_rate": 4.729196740308696e-05,
+      "loss": 3.6771,
+      "step": 15850
+    },
+    {
+      "epoch": 5.268411896280341,
+      "grad_norm": 0.47265625,
+      "learning_rate": 4.7271690744172745e-05,
+      "loss": 3.6752,
+      "step": 15900
+    },
+    {
+      "epoch": 5.2849805318532015,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.725134283666607e-05,
+      "loss": 3.6766,
+      "step": 15950
+    },
+    {
+      "epoch": 5.301549167426063,
+      "grad_norm": 0.486328125,
+      "learning_rate": 4.723092374566123e-05,
+      "loss": 3.6753,
+      "step": 16000
+    },
+    {
+      "epoch": 5.318117802998923,
+      "grad_norm": 0.478515625,
+      "learning_rate": 4.721043353648021e-05,
+      "loss": 3.6728,
+      "step": 16050
+    },
+    {
+      "epoch": 5.334686438571784,
+      "grad_norm": 0.4921875,
+      "learning_rate": 4.718987227467256e-05,
+      "loss": 3.6722,
+      "step": 16100
+    },
+    {
+      "epoch": 5.351255074144644,
+      "grad_norm": 0.486328125,
+      "learning_rate": 4.716924002601507e-05,
+      "loss": 3.6674,
+      "step": 16150
+    },
+    {
+      "epoch": 5.367823709717505,
+      "grad_norm": 0.478515625,
+      "learning_rate": 4.7148536856511655e-05,
+      "loss": 3.6766,
+      "step": 16200
+    },
+    {
+      "epoch": 5.384392345290365,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.712776283239311e-05,
+      "loss": 3.6643,
+      "step": 16250
+    },
+    {
+      "epoch": 5.400960980863226,
+      "grad_norm": 0.49609375,
+      "learning_rate": 4.71069180201169e-05,
+      "loss": 3.6655,
+      "step": 16300
+    },
+    {
+      "epoch": 5.417529616436086,
+      "grad_norm": 0.5,
+      "learning_rate": 4.708600248636693e-05,
+      "loss": 3.6734,
+      "step": 16350
+    },
+    {
+      "epoch": 5.434098252008947,
+      "grad_norm": 0.498046875,
+      "learning_rate": 4.7065016298053365e-05,
+      "loss": 3.6684,
+      "step": 16400
+    },
+    {
+      "epoch": 5.450666887581807,
+      "grad_norm": 0.47265625,
+      "learning_rate": 4.7043959522312384e-05,
+      "loss": 3.6661,
+      "step": 16450
+    },
+    {
+      "epoch": 5.4672355231546685,
+      "grad_norm": 0.462890625,
+      "learning_rate": 4.7022832226505996e-05,
+      "loss": 3.6706,
+      "step": 16500
+    },
+    {
+      "epoch": 5.483804158727529,
+      "grad_norm": 0.478515625,
+      "learning_rate": 4.700163447822179e-05,
+      "loss": 3.6701,
+      "step": 16550
+    },
+    {
+      "epoch": 5.5003727943003895,
+      "grad_norm": 0.5,
+      "learning_rate": 4.6980366345272756e-05,
+      "loss": 3.6724,
+      "step": 16600
+    },
+    {
+      "epoch": 5.51694142987325,
+      "grad_norm": 0.46875,
+      "learning_rate": 4.6959027895697026e-05,
+      "loss": 3.6691,
+      "step": 16650
+    },
+    {
+      "epoch": 5.5335100654461105,
+      "grad_norm": 0.486328125,
+      "learning_rate": 4.693761919775771e-05,
+      "loss": 3.6729,
+      "step": 16700
+    },
+    {
+      "epoch": 5.550078701018971,
+      "grad_norm": 0.484375,
+      "learning_rate": 4.691614031994261e-05,
+      "loss": 3.6766,
+      "step": 16750
+    },
+    {
+      "epoch": 5.566647336591831,
+      "grad_norm": 0.458984375,
+      "learning_rate": 4.6894591330964076e-05,
+      "loss": 3.6744,
+      "step": 16800
+    },
+    {
+      "epoch": 5.583215972164692,
+      "grad_norm": 0.46875,
+      "learning_rate": 4.6872972299758726e-05,
+      "loss": 3.6675,
+      "step": 16850
+    },
+    {
+      "epoch": 5.599784607737552,
+      "grad_norm": 0.498046875,
+      "learning_rate": 4.685128329548725e-05,
+      "loss": 3.6669,
+      "step": 16900
+    },
+    {
+      "epoch": 5.616353243310414,
+      "grad_norm": 0.46875,
+      "learning_rate": 4.6829524387534184e-05,
+      "loss": 3.6688,
+      "step": 16950
+    },
+    {
+      "epoch": 5.632921878883274,
+      "grad_norm": 0.474609375,
+      "learning_rate": 4.6807695645507704e-05,
+      "loss": 3.6708,
+      "step": 17000
+    },
+    {
+      "epoch": 5.649490514456135,
+      "grad_norm": 0.4765625,
+      "learning_rate": 4.6785797139239376e-05,
+      "loss": 3.669,
+      "step": 17050
+    },
+    {
+      "epoch": 5.666059150028995,
+      "grad_norm": 0.494140625,
+      "learning_rate": 4.676382893878395e-05,
+      "loss": 3.6652,
+      "step": 17100
+    },
+    {
+      "epoch": 5.682627785601856,
+      "grad_norm": 0.482421875,
+      "learning_rate": 4.674179111441913e-05,
+      "loss": 3.6658,
+      "step": 17150
+    },
+    {
+      "epoch": 5.699196421174716,
+      "grad_norm": 0.48828125,
+      "learning_rate": 4.671968373664536e-05,
+      "loss": 3.6668,
+      "step": 17200
+    },
+    {
+      "epoch": 5.715765056747577,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.6697506876185586e-05,
+      "loss": 3.6721,
+      "step": 17250
+    },
+    {
+      "epoch": 5.732333692320437,
+      "grad_norm": 0.48828125,
+      "learning_rate": 4.667526060398503e-05,
+      "loss": 3.6696,
+      "step": 17300
+    },
+    {
+      "epoch": 5.748902327893298,
+      "grad_norm": 0.474609375,
+      "learning_rate": 4.665294499121097e-05,
+      "loss": 3.6714,
+      "step": 17350
+    },
+    {
+      "epoch": 5.765470963466159,
+      "grad_norm": 0.494140625,
+      "learning_rate": 4.6630560109252496e-05,
+      "loss": 3.665,
+      "step": 17400
+    },
+    {
+      "epoch": 5.782039599039019,
+      "grad_norm": 0.478515625,
+      "learning_rate": 4.660810602972032e-05,
+      "loss": 3.6705,
+      "step": 17450
+    },
+    {
+      "epoch": 5.79860823461188,
+      "grad_norm": 0.466796875,
+      "learning_rate": 4.658558282444651e-05,
+      "loss": 3.6658,
+      "step": 17500
+    },
+    {
+      "epoch": 5.81517687018474,
+      "grad_norm": 0.474609375,
+      "learning_rate": 4.6562990565484265e-05,
+      "loss": 3.6624,
+      "step": 17550
+    },
+    {
+      "epoch": 5.831745505757601,
+      "grad_norm": 0.46484375,
+      "learning_rate": 4.654032932510769e-05,
+      "loss": 3.6718,
+      "step": 17600
+    },
+    {
+      "epoch": 5.848314141330461,
+      "grad_norm": 0.47265625,
+      "learning_rate": 4.651759917581159e-05,
+      "loss": 3.6682,
+      "step": 17650
+    },
+    {
+      "epoch": 5.864882776903322,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.649480019031117e-05,
+      "loss": 3.6631,
+      "step": 17700
+    },
+    {
+      "epoch": 5.881451412476182,
+      "grad_norm": 0.48828125,
+      "learning_rate": 4.647193244154189e-05,
+      "loss": 3.6697,
+      "step": 17750
+    },
+    {
+      "epoch": 5.898020048049043,
+      "grad_norm": 0.45703125,
+      "learning_rate": 4.644899600265917e-05,
+      "loss": 3.6689,
+      "step": 17800
+    },
+    {
+      "epoch": 5.914588683621904,
+      "grad_norm": 0.5,
+      "learning_rate": 4.6425990947038175e-05,
+      "loss": 3.6726,
+      "step": 17850
+    },
+    {
+      "epoch": 5.931157319194765,
+      "grad_norm": 0.46875,
+      "learning_rate": 4.640291734827357e-05,
+      "loss": 3.6715,
+      "step": 17900
+    },
+    {
+      "epoch": 5.947725954767625,
+      "grad_norm": 0.478515625,
+      "learning_rate": 4.637977528017931e-05,
+      "loss": 3.67,
+      "step": 17950
+    },
+    {
+      "epoch": 5.964294590340486,
+      "grad_norm": 0.46484375,
+      "learning_rate": 4.63565648167884e-05,
+      "loss": 3.6715,
+      "step": 18000
+    },
+    {
+      "epoch": 5.980863225913346,
+      "grad_norm": 0.4765625,
+      "learning_rate": 4.6333286032352594e-05,
+      "loss": 3.6649,
+      "step": 18050
+    },
+    {
+      "epoch": 5.9974318614862066,
+      "grad_norm": 0.482421875,
+      "learning_rate": 4.630993900134227e-05,
+      "loss": 3.6725,
+      "step": 18100
+    },
+    {
+      "epoch": 6.013917653881203,
+      "grad_norm": 0.53515625,
+      "learning_rate": 4.628652379844609e-05,
+      "loss": 3.6574,
+      "step": 18150
+    },
+    {
+      "epoch": 6.030486289454063,
+      "grad_norm": 0.53125,
+      "learning_rate": 4.626304049857081e-05,
+      "loss": 3.6601,
+      "step": 18200
+    },
+    {
+      "epoch": 6.047054925026924,
+      "grad_norm": 0.5703125,
+      "learning_rate": 4.623948917684106e-05,
+      "loss": 3.67,
+      "step": 18250
+    },
+    {
+      "epoch": 6.063623560599785,
+      "grad_norm": 0.5625,
+      "learning_rate": 4.621586990859904e-05,
+      "loss": 3.6611,
+      "step": 18300
+    },
+    {
+      "epoch": 6.0801921961726455,
+      "grad_norm": 0.53125,
+      "learning_rate": 4.6192182769404335e-05,
+      "loss": 3.6608,
+      "step": 18350
+    },
+    {
+      "epoch": 6.096760831745506,
+      "grad_norm": 0.53125,
+      "learning_rate": 4.616842783503365e-05,
+      "loss": 3.6631,
+      "step": 18400
+    },
+    {
+      "epoch": 6.1133294673183665,
+      "grad_norm": 0.53125,
+      "learning_rate": 4.614460518148057e-05,
+      "loss": 3.657,
+      "step": 18450
+    },
+    {
+      "epoch": 6.129898102891227,
+      "grad_norm": 0.5234375,
+      "learning_rate": 4.6120714884955316e-05,
+      "loss": 3.6635,
+      "step": 18500
+    },
+    {
+      "epoch": 6.146466738464087,
+      "grad_norm": 0.546875,
+      "learning_rate": 4.609675702188452e-05,
+      "loss": 3.6595,
+      "step": 18550
+    },
+    {
+      "epoch": 6.163035374036948,
+      "grad_norm": 0.55078125,
+      "learning_rate": 4.6072731668910926e-05,
+      "loss": 3.6676,
+      "step": 18600
+    },
+    {
+      "epoch": 6.179604009609808,
+      "grad_norm": 0.5625,
+      "learning_rate": 4.604863890289322e-05,
+      "loss": 3.6611,
+      "step": 18650
+    },
+    {
+      "epoch": 6.196172645182669,
+      "grad_norm": 0.546875,
+      "learning_rate": 4.6024478800905735e-05,
+      "loss": 3.6679,
+      "step": 18700
+    },
+    {
+      "epoch": 6.21274128075553,
+      "grad_norm": 0.54296875,
+      "learning_rate": 4.600025144023822e-05,
+      "loss": 3.6612,
+      "step": 18750
+    },
+    {
+      "epoch": 6.229309916328391,
+      "grad_norm": 0.56640625,
+      "learning_rate": 4.597595689839556e-05,
+      "loss": 3.6664,
+      "step": 18800
+    },
+    {
+      "epoch": 6.245878551901251,
+      "grad_norm": 0.546875,
+      "learning_rate": 4.59515952530976e-05,
+      "loss": 3.6644,
+      "step": 18850
+    },
+    {
+      "epoch": 6.262447187474112,
+      "grad_norm": 0.52734375,
+      "learning_rate": 4.592716658227884e-05,
+      "loss": 3.6656,
+      "step": 18900
+    },
+    {
+      "epoch": 6.279015823046972,
+      "grad_norm": 0.53515625,
+      "learning_rate": 4.5902670964088165e-05,
+      "loss": 3.6643,
+      "step": 18950
+    },
+    {
+      "epoch": 6.295584458619833,
+      "grad_norm": 0.55859375,
+      "learning_rate": 4.587810847688868e-05,
+      "loss": 3.6589,
+      "step": 19000
+    },
+    {
+      "epoch": 6.312153094192693,
+      "grad_norm": 0.55078125,
+      "learning_rate": 4.585347919925737e-05,
+      "loss": 3.6601,
+      "step": 19050
+    },
+    {
+      "epoch": 6.328721729765554,
+      "grad_norm": 0.54296875,
+      "learning_rate": 4.582878320998491e-05,
+      "loss": 3.662,
+      "step": 19100
+    },
+    {
+      "epoch": 6.345290365338414,
+      "grad_norm": 0.52734375,
+      "learning_rate": 4.580402058807539e-05,
+      "loss": 3.6678,
+      "step": 19150
+    },
+    {
+      "epoch": 6.3618590009112745,
+      "grad_norm": 0.51953125,
+      "learning_rate": 4.577919141274604e-05,
+      "loss": 3.6652,
+      "step": 19200
+    },
+    {
+      "epoch": 6.378427636484136,
+      "grad_norm": 0.53515625,
+      "learning_rate": 4.575429576342701e-05,
+      "loss": 3.6591,
+      "step": 19250
+    },
+    {
+      "epoch": 6.394996272056996,
+      "grad_norm": 0.55078125,
+      "learning_rate": 4.5729333719761124e-05,
+      "loss": 3.6592,
+      "step": 19300
+    },
+    {
+      "epoch": 6.411564907629857,
+      "grad_norm": 0.546875,
+      "learning_rate": 4.5704305361603585e-05,
+      "loss": 3.66,
+      "step": 19350
+    },
+    {
+      "epoch": 6.428133543202717,
+      "grad_norm": 0.52734375,
+      "learning_rate": 4.567921076902173e-05,
+      "loss": 3.6641,
+      "step": 19400
+    },
+    {
+      "epoch": 6.444702178775578,
+      "grad_norm": 0.54296875,
+      "learning_rate": 4.5654050022294805e-05,
+      "loss": 3.6676,
+      "step": 19450
+    },
+    {
+      "epoch": 6.461270814348438,
+      "grad_norm": 0.55859375,
+      "learning_rate": 4.562882320191368e-05,
+      "loss": 3.6594,
+      "step": 19500
+    },
+    {
+      "epoch": 6.477839449921299,
+      "grad_norm": 0.52734375,
+      "learning_rate": 4.560353038858062e-05,
+      "loss": 3.6554,
+      "step": 19550
+    },
+    {
+      "epoch": 6.494408085494159,
+      "grad_norm": 0.52734375,
+      "learning_rate": 4.557817166320896e-05,
+      "loss": 3.668,
+      "step": 19600
+    },
+    {
+      "epoch": 6.51097672106702,
+      "grad_norm": 0.546875,
+      "learning_rate": 4.5552747106922924e-05,
+      "loss": 3.6629,
+      "step": 19650
+    },
+    {
+      "epoch": 6.52754535663988,
+      "grad_norm": 0.515625,
+      "learning_rate": 4.552725680105733e-05,
+      "loss": 3.666,
+      "step": 19700
+    },
+    {
+      "epoch": 6.544113992212742,
+      "grad_norm": 0.5390625,
+      "learning_rate": 4.550170082715733e-05,
+      "loss": 3.6651,
+      "step": 19750
+    },
+    {
+      "epoch": 6.560682627785602,
+      "grad_norm": 0.5546875,
+      "learning_rate": 4.5476079266978164e-05,
+      "loss": 3.6653,
+      "step": 19800
+    },
+    {
+      "epoch": 6.5772512633584626,
+      "grad_norm": 0.5546875,
+      "learning_rate": 4.545039220248486e-05,
+      "loss": 3.6644,
+      "step": 19850
+    },
+    {
+      "epoch": 6.593819898931323,
+      "grad_norm": 0.546875,
+      "learning_rate": 4.542463971585201e-05,
+      "loss": 3.6624,
+      "step": 19900
+    },
+    {
+      "epoch": 6.6103885345041835,
+      "grad_norm": 0.53125,
+      "learning_rate": 4.539882188946352e-05,
+      "loss": 3.6577,
+      "step": 19950
+    },
+    {
+      "epoch": 6.626957170077044,
+      "grad_norm": 0.53515625,
+      "learning_rate": 4.537293880591229e-05,
+      "loss": 3.6674,
+      "step": 20000
+    },
+    {
+      "epoch": 6.626957170077044,
+      "eval_loss": 3.6582796573638916,
+      "eval_runtime": 7.9393,
+      "eval_samples_per_second": 122.05,
+      "eval_steps_per_second": 2.015,
+      "step": 20000
+    },
+    {
+      "epoch": 6.6435258056499045,
+      "grad_norm": 0.54296875,
+      "learning_rate": 4.5346990547999984e-05,
+      "loss": 3.6666,
+      "step": 20050
+    },
+    {
+      "epoch": 6.660094441222765,
+      "grad_norm": 0.56640625,
+      "learning_rate": 4.532097719873679e-05,
+      "loss": 3.667,
+      "step": 20100
+    },
+    {
+      "epoch": 6.676663076795625,
+      "grad_norm": 0.5625,
+      "learning_rate": 4.52948988413411e-05,
+      "loss": 3.6659,
+      "step": 20150
+    },
+    {
+      "epoch": 6.693231712368487,
+      "grad_norm": 0.54296875,
+      "learning_rate": 4.526875555923929e-05,
+      "loss": 3.6622,
+      "step": 20200
+    },
+    {
+      "epoch": 6.709800347941347,
+      "grad_norm": 0.5625,
+      "learning_rate": 4.524254743606542e-05,
+      "loss": 3.6651,
+      "step": 20250
+    },
+    {
+      "epoch": 6.726368983514208,
+      "grad_norm": 0.55078125,
+      "learning_rate": 4.5216274555660984e-05,
+      "loss": 3.6593,
+      "step": 20300
+    },
+    {
+      "epoch": 6.742937619087068,
+      "grad_norm": 0.52734375,
+      "learning_rate": 4.518993700207464e-05,
+      "loss": 3.6625,
+      "step": 20350
+    },
+    {
+      "epoch": 6.759506254659929,
+      "grad_norm": 0.546875,
+      "learning_rate": 4.516353485956196e-05,
+      "loss": 3.6543,
+      "step": 20400
+    },
+    {
+      "epoch": 6.776074890232789,
+      "grad_norm": 0.5390625,
+      "learning_rate": 4.51370682125851e-05,
+      "loss": 3.6597,
+      "step": 20450
+    },
+    {
+      "epoch": 6.79264352580565,
+      "grad_norm": 0.55859375,
+      "learning_rate": 4.511053714581259e-05,
+      "loss": 3.6657,
+      "step": 20500
+    },
+    {
+      "epoch": 6.80921216137851,
+      "grad_norm": 0.53125,
+      "learning_rate": 4.508394174411907e-05,
+      "loss": 3.6664,
+      "step": 20550
+    },
+    {
+      "epoch": 6.825780796951371,
+      "grad_norm": 0.55078125,
+      "learning_rate": 4.5057282092584933e-05,
+      "loss": 3.6632,
+      "step": 20600
+    },
+    {
+      "epoch": 6.842349432524232,
+      "grad_norm": 0.53515625,
+      "learning_rate": 4.503055827649617e-05,
+      "loss": 3.6569,
+      "step": 20650
+    },
+    {
+      "epoch": 6.8589180680970925,
+      "grad_norm": 0.54296875,
+      "learning_rate": 4.5003770381344e-05,
+      "loss": 3.6661,
+      "step": 20700
+    },
+    {
+      "epoch": 6.875486703669953,
+      "grad_norm": 0.5390625,
+      "learning_rate": 4.4976918492824655e-05,
+      "loss": 3.6716,
+      "step": 20750
+    },
+    {
+      "epoch": 6.892055339242813,
+      "grad_norm": 0.54296875,
+      "learning_rate": 4.495000269683907e-05,
+      "loss": 3.658,
+      "step": 20800
+    },
+    {
+      "epoch": 6.908623974815674,
+      "grad_norm": 0.5625,
+      "learning_rate": 4.4923023079492646e-05,
+      "loss": 3.6617,
+      "step": 20850
+    },
+    {
+      "epoch": 6.925192610388534,
+      "grad_norm": 0.53515625,
+      "learning_rate": 4.489597972709492e-05,
+      "loss": 3.6573,
+      "step": 20900
+    },
+    {
+      "epoch": 6.941761245961395,
+      "grad_norm": 0.5390625,
+      "learning_rate": 4.4868872726159364e-05,
+      "loss": 3.6628,
+      "step": 20950
+    },
+    {
+      "epoch": 6.958329881534255,
+      "grad_norm": 0.53125,
+      "learning_rate": 4.484170216340303e-05,
+      "loss": 3.6677,
+      "step": 21000
+    },
+    {
+      "epoch": 6.974898517107116,
+      "grad_norm": 0.5546875,
+      "learning_rate": 4.481446812574633e-05,
+      "loss": 3.6624,
+      "step": 21050
+    },
+    {
+      "epoch": 6.991467152679977,
+      "grad_norm": 0.53515625,
+      "learning_rate": 4.478717070031272e-05,
+      "loss": 3.6585,
+      "step": 21100
+    },
+    {
+      "epoch": 7.007952945074973,
+      "grad_norm": 0.62890625,
+      "learning_rate": 4.475980997442846e-05,
+      "loss": 3.653,
+      "step": 21150
+    },
+    {
+      "epoch": 7.024521580647834,
+      "grad_norm": 0.62109375,
+      "learning_rate": 4.4732386035622276e-05,
+      "loss": 3.6523,
+      "step": 21200
+    },
+    {
+      "epoch": 7.041090216220694,
+      "grad_norm": 0.62890625,
+      "learning_rate": 4.470489897162515e-05,
+      "loss": 3.6568,
+      "step": 21250
+    },
+    {
+      "epoch": 7.057658851793555,
+      "grad_norm": 0.65234375,
+      "learning_rate": 4.467734887037e-05,
+      "loss": 3.6541,
+      "step": 21300
+    },
+    {
+      "epoch": 7.074227487366415,
+      "grad_norm": 0.62109375,
+      "learning_rate": 4.464973581999138e-05,
+      "loss": 3.6499,
+      "step": 21350
+    },
+    {
+      "epoch": 7.090796122939276,
+      "grad_norm": 0.64453125,
+      "learning_rate": 4.462205990882526e-05,
+      "loss": 3.6529,
+      "step": 21400
+    },
+    {
+      "epoch": 7.107364758512136,
+      "grad_norm": 0.640625,
+      "learning_rate": 4.4594321225408686e-05,
+      "loss": 3.6641,
+      "step": 21450
+    },
+    {
+      "epoch": 7.123933394084997,
+      "grad_norm": 0.640625,
+      "learning_rate": 4.4566519858479506e-05,
+      "loss": 3.6618,
+      "step": 21500
+    },
+    {
+      "epoch": 7.140502029657858,
+      "grad_norm": 0.66796875,
+      "learning_rate": 4.45386558969761e-05,
+      "loss": 3.6594,
+      "step": 21550
+    },
+    {
+      "epoch": 7.1570706652307186,
+      "grad_norm": 0.64453125,
+      "learning_rate": 4.451072943003712e-05,
+      "loss": 3.6543,
+      "step": 21600
+    },
+    {
+      "epoch": 7.173639300803579,
+      "grad_norm": 0.65234375,
+      "learning_rate": 4.4482740547001154e-05,
+      "loss": 3.6566,
+      "step": 21650
+    },
+    {
+      "epoch": 7.1902079363764395,
+      "grad_norm": 0.671875,
+      "learning_rate": 4.445468933740645e-05,
+      "loss": 3.6592,
+      "step": 21700
+    },
+    {
+      "epoch": 7.2067765719493,
+      "grad_norm": 0.6484375,
+      "learning_rate": 4.4426575890990675e-05,
+      "loss": 3.6539,
+      "step": 21750
+    },
+    {
+      "epoch": 7.2233452075221605,
+      "grad_norm": 0.63671875,
+      "learning_rate": 4.439840029769056e-05,
+      "loss": 3.6545,
+      "step": 21800
+    },
+    {
+      "epoch": 7.239913843095021,
+      "grad_norm": 0.64453125,
+      "learning_rate": 4.437016264764169e-05,
+      "loss": 3.6614,
+      "step": 21850
+    },
+    {
+      "epoch": 7.256482478667881,
+      "grad_norm": 0.62109375,
+      "learning_rate": 4.434186303117814e-05,
+      "loss": 3.6579,
+      "step": 21900
+    },
+    {
+      "epoch": 7.273051114240742,
+      "grad_norm": 0.60546875,
+      "learning_rate": 4.431350153883223e-05,
+      "loss": 3.6529,
+      "step": 21950
+    },
+    {
+      "epoch": 7.289619749813603,
+      "grad_norm": 0.66015625,
+      "learning_rate": 4.4285078261334215e-05,
+      "loss": 3.6568,
+      "step": 22000
+    },
+    {
+      "epoch": 7.306188385386464,
+      "grad_norm": 0.625,
+      "learning_rate": 4.425659328961203e-05,
+      "loss": 3.6534,
+      "step": 22050
+    },
+    {
+      "epoch": 7.322757020959324,
+      "grad_norm": 0.65234375,
+      "learning_rate": 4.422804671479096e-05,
+      "loss": 3.653,
+      "step": 22100
+    },
+    {
+      "epoch": 7.339325656532185,
+      "grad_norm": 0.6484375,
+      "learning_rate": 4.4199438628193345e-05,
+      "loss": 3.6606,
+      "step": 22150
+    },
+    {
+      "epoch": 7.355894292105045,
+      "grad_norm": 0.65234375,
+      "learning_rate": 4.417076912133834e-05,
+      "loss": 3.6566,
+      "step": 22200
+    },
+    {
+      "epoch": 7.372462927677906,
+      "grad_norm": 0.63671875,
+      "learning_rate": 4.414203828594156e-05,
+      "loss": 3.664,
+      "step": 22250
+    },
+    {
+      "epoch": 7.389031563250766,
+      "grad_norm": 0.6640625,
+      "learning_rate": 4.4113246213914826e-05,
+      "loss": 3.657,
+      "step": 22300
+    },
+    {
+      "epoch": 7.405600198823627,
+      "grad_norm": 0.640625,
+      "learning_rate": 4.408439299736585e-05,
+      "loss": 3.6524,
+      "step": 22350
+    },
+    {
+      "epoch": 7.422168834396487,
+      "grad_norm": 0.6328125,
+      "learning_rate": 4.405547872859796e-05,
+      "loss": 3.655,
+      "step": 22400
+    },
+    {
+      "epoch": 7.4387374699693485,
+      "grad_norm": 0.66015625,
+      "learning_rate": 4.4026503500109785e-05,
+      "loss": 3.6489,
+      "step": 22450
+    },
+    {
+      "epoch": 7.455306105542209,
+      "grad_norm": 0.63671875,
+      "learning_rate": 4.399746740459498e-05,
+      "loss": 3.6565,
+      "step": 22500
+    },
+    {
+      "epoch": 7.471874741115069,
+      "grad_norm": 0.65625,
+      "learning_rate": 4.39683705349419e-05,
+      "loss": 3.6606,
+      "step": 22550
+    },
+    {
+      "epoch": 7.48844337668793,
+      "grad_norm": 0.640625,
+      "learning_rate": 4.393921298423334e-05,
+      "loss": 3.6599,
+      "step": 22600
+    },
+    {
+      "epoch": 7.50501201226079,
+      "grad_norm": 0.6171875,
+      "learning_rate": 4.390999484574619e-05,
+      "loss": 3.6506,
+      "step": 22650
+    },
+    {
+      "epoch": 7.521580647833651,
+      "grad_norm": 0.6328125,
+      "learning_rate": 4.38807162129512e-05,
+      "loss": 3.6527,
+      "step": 22700
+    },
+    {
+      "epoch": 7.538149283406511,
+      "grad_norm": 0.66015625,
+      "learning_rate": 4.385137717951263e-05,
+      "loss": 3.6602,
+      "step": 22750
+    },
+    {
+      "epoch": 7.554717918979372,
+      "grad_norm": 0.6484375,
+      "learning_rate": 4.3821977839287954e-05,
+      "loss": 3.6525,
+      "step": 22800
+    },
+    {
+      "epoch": 7.571286554552232,
+      "grad_norm": 0.6484375,
+      "learning_rate": 4.3792518286327576e-05,
+      "loss": 3.6623,
+      "step": 22850
+    },
+    {
+      "epoch": 7.587855190125094,
+      "grad_norm": 0.6171875,
+      "learning_rate": 4.376299861487454e-05,
+      "loss": 3.6546,
+      "step": 22900
+    },
+    {
+      "epoch": 7.604423825697953,
+      "grad_norm": 0.63671875,
+      "learning_rate": 4.37334189193642e-05,
+      "loss": 3.6664,
+      "step": 22950
+    },
+    {
+      "epoch": 7.620992461270815,
+      "grad_norm": 0.65625,
+      "learning_rate": 4.3703779294423925e-05,
+      "loss": 3.6585,
+      "step": 23000
+    },
+    {
+      "epoch": 7.637561096843675,
+      "grad_norm": 0.64453125,
+      "learning_rate": 4.367407983487282e-05,
+      "loss": 3.6566,
+      "step": 23050
+    },
+    {
+      "epoch": 7.654129732416536,
+      "grad_norm": 0.6484375,
+      "learning_rate": 4.3644320635721374e-05,
+      "loss": 3.6595,
+      "step": 23100
+    },
+    {
+      "epoch": 7.670698367989396,
+      "grad_norm": 0.66015625,
+      "learning_rate": 4.3614501792171237e-05,
+      "loss": 3.6528,
+      "step": 23150
+    },
+    {
+      "epoch": 7.687267003562257,
+      "grad_norm": 0.640625,
+      "learning_rate": 4.358462339961481e-05,
+      "loss": 3.657,
+      "step": 23200
+    },
+    {
+      "epoch": 7.703835639135117,
+      "grad_norm": 0.6484375,
+      "learning_rate": 4.355468555363504e-05,
+      "loss": 3.6571,
+      "step": 23250
+    },
+    {
+      "epoch": 7.7204042747079775,
+      "grad_norm": 0.640625,
+      "learning_rate": 4.3524688350005046e-05,
+      "loss": 3.6558,
+      "step": 23300
+    },
+    {
+      "epoch": 7.736972910280838,
+      "grad_norm": 0.6875,
+      "learning_rate": 4.349463188468783e-05,
+      "loss": 3.6608,
+      "step": 23350
+    },
+    {
+      "epoch": 7.7535415458536985,
+      "grad_norm": 0.6484375,
+      "learning_rate": 4.3464516253836e-05,
+      "loss": 3.6621,
+      "step": 23400
+    },
+    {
+      "epoch": 7.77011018142656,
+      "grad_norm": 0.64453125,
+      "learning_rate": 4.343434155379142e-05,
+      "loss": 3.6537,
+      "step": 23450
+    },
+    {
+      "epoch": 7.78667881699942,
+      "grad_norm": 0.6484375,
+      "learning_rate": 4.340410788108492e-05,
+      "loss": 3.6514,
+      "step": 23500
+    },
+    {
+      "epoch": 7.803247452572281,
+      "grad_norm": 0.67578125,
+      "learning_rate": 4.3373815332436e-05,
+      "loss": 3.6655,
+      "step": 23550
+    },
+    {
+      "epoch": 7.819816088145141,
+      "grad_norm": 0.64453125,
+      "learning_rate": 4.334346400475249e-05,
+      "loss": 3.6557,
+      "step": 23600
+    },
+    {
+      "epoch": 7.836384723718002,
+      "grad_norm": 0.6328125,
+      "learning_rate": 4.3313053995130283e-05,
+      "loss": 3.6564,
+      "step": 23650
+    },
+    {
+      "epoch": 7.852953359290862,
+      "grad_norm": 0.66015625,
+      "learning_rate": 4.328258540085297e-05,
+      "loss": 3.6568,
+      "step": 23700
+    },
+    {
+      "epoch": 7.869521994863723,
+      "grad_norm": 0.6484375,
+      "learning_rate": 4.3252058319391585e-05,
+      "loss": 3.6553,
+      "step": 23750
+    },
+    {
+      "epoch": 7.886090630436583,
+      "grad_norm": 0.671875,
+      "learning_rate": 4.322147284840424e-05,
+      "loss": 3.6548,
+      "step": 23800
+    },
+    {
+      "epoch": 7.902659266009444,
+      "grad_norm": 0.65625,
+      "learning_rate": 4.3190829085735875e-05,
+      "loss": 3.6557,
+      "step": 23850
+    },
+    {
+      "epoch": 7.919227901582305,
+      "grad_norm": 0.6484375,
+      "learning_rate": 4.3160127129417874e-05,
+      "loss": 3.6587,
+      "step": 23900
+    },
+    {
+      "epoch": 7.9357965371551655,
+      "grad_norm": 0.6328125,
+      "learning_rate": 4.3129367077667796e-05,
+      "loss": 3.6563,
+      "step": 23950
+    },
+    {
+      "epoch": 7.952365172728026,
+      "grad_norm": 0.69140625,
+      "learning_rate": 4.309854902888904e-05,
+      "loss": 3.6571,
+      "step": 24000
+    },
+    {
+      "epoch": 7.9689338083008865,
+      "grad_norm": 0.62890625,
+      "learning_rate": 4.30676730816706e-05,
+      "loss": 3.6603,
+      "step": 24050
+    },
+    {
+      "epoch": 7.985502443873747,
+      "grad_norm": 0.63671875,
+      "learning_rate": 4.3036739334786615e-05,
+      "loss": 3.6453,
+      "step": 24100
+    },
+    {
+      "epoch": 8.001988236268744,
+      "grad_norm": 0.828125,
+      "learning_rate": 4.3005747887196155e-05,
+      "loss": 3.6534,
+      "step": 24150
+    },
+    {
+      "epoch": 8.018556871841604,
+      "grad_norm": 0.84375,
+      "learning_rate": 4.297469883804291e-05,
+      "loss": 3.65,
+      "step": 24200
+    },
+    {
+      "epoch": 8.035125507414465,
+      "grad_norm": 0.8515625,
+      "learning_rate": 4.2943592286654785e-05,
+      "loss": 3.6524,
+      "step": 24250
+    },
+    {
+      "epoch": 8.051694142987325,
+      "grad_norm": 0.8671875,
+      "learning_rate": 4.291242833254369e-05,
+      "loss": 3.6519,
+      "step": 24300
+    },
+    {
+      "epoch": 8.068262778560186,
+      "grad_norm": 0.859375,
+      "learning_rate": 4.288120707540514e-05,
+      "loss": 3.653,
+      "step": 24350
+    },
+    {
+      "epoch": 8.084831414133046,
+      "grad_norm": 0.8203125,
+      "learning_rate": 4.284992861511797e-05,
+      "loss": 3.6516,
+      "step": 24400
+    },
+    {
+      "epoch": 8.101400049705907,
+      "grad_norm": 0.90234375,
+      "learning_rate": 4.281859305174403e-05,
+      "loss": 3.6507,
+      "step": 24450
+    },
+    {
+      "epoch": 8.117968685278766,
+      "grad_norm": 0.85546875,
+      "learning_rate": 4.278720048552781e-05,
+      "loss": 3.6543,
+      "step": 24500
+    },
+    {
+      "epoch": 8.134537320851628,
+      "grad_norm": 0.84375,
+      "learning_rate": 4.275575101689621e-05,
+      "loss": 3.6556,
+      "step": 24550
+    },
+    {
+      "epoch": 8.15110595642449,
+      "grad_norm": 0.83203125,
+      "learning_rate": 4.2724244746458105e-05,
+      "loss": 3.6472,
+      "step": 24600
+    },
+    {
+      "epoch": 8.167674591997349,
+      "grad_norm": 0.82421875,
+      "learning_rate": 4.269268177500413e-05,
+      "loss": 3.6512,
+      "step": 24650
+    },
+    {
+      "epoch": 8.18424322757021,
+      "grad_norm": 0.84375,
+      "learning_rate": 4.266106220350628e-05,
+      "loss": 3.653,
+      "step": 24700
+    },
+    {
+      "epoch": 8.20081186314307,
+      "grad_norm": 0.84375,
+      "learning_rate": 4.262938613311763e-05,
+      "loss": 3.6573,
+      "step": 24750
+    },
+    {
+      "epoch": 8.217380498715931,
+      "grad_norm": 0.81640625,
+      "learning_rate": 4.259765366517199e-05,
+      "loss": 3.6489,
+      "step": 24800
+    },
+    {
+      "epoch": 8.23394913428879,
+      "grad_norm": 0.78515625,
+      "learning_rate": 4.2565864901183615e-05,
+      "loss": 3.6485,
+      "step": 24850
+    },
+    {
+      "epoch": 8.250517769861652,
+      "grad_norm": 0.83984375,
+      "learning_rate": 4.2534019942846815e-05,
+      "loss": 3.6548,
+      "step": 24900
+    },
+    {
+      "epoch": 8.267086405434512,
+      "grad_norm": 0.859375,
+      "learning_rate": 4.250211889203571e-05,
+      "loss": 3.6527,
+      "step": 24950
+    },
+    {
+      "epoch": 8.283655041007373,
+      "grad_norm": 0.828125,
+      "learning_rate": 4.247016185080382e-05,
+      "loss": 3.6522,
+      "step": 25000
+    },
+    {
+      "epoch": 8.283655041007373,
+      "eval_loss": 3.6497414112091064,
+      "eval_runtime": 8.0787,
+      "eval_samples_per_second": 119.945,
+      "eval_steps_per_second": 1.981,
+      "step": 25000
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 90540,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 30,
+  "save_steps": 5000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.379230310816285e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-25000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28bc9829b8c8fa8bf181924e8f4aa71e88dec472de000157e8bb54e4440b7211
+size 5841