* 2023/03/20 Update the model weight and config files such that it can be loaded via Huggingface's official GPT-NeoX implementation.

README.md

@@ -22,17 +22,15 @@ inference: false
 
 This repository provides a small-sized Japanese GPT-NeoX model. The model was trained using code based on [EleutherAI/gpt-neox](https://github.com/EleutherAI/gpt-neox).
 
-# How to use the model
+# Update log
+* 2023/03/20 Update the model weight and config files such that it can be loaded via Huggingface's official GPT-NeoX implementation.
 
-*NOTE:*
-* Use `T5Tokenizer` to load its corresponding tokenizer.
-* The files for modeling and configuration are not in the Transformers library yet. In order to load the model, use files from [this PR in EleutherAI/gpt-neox](https://github.com/EleutherAI/gpt-neox/pull/480).
+# How to use the model
 
 ~~~~
-from transformers import T5Tokenizer
-from modeling_gpt_neox import GPTNeoXForCausalLM
+from transformers import AutoTokenizer, AutoModelForCausalLM
 
-tokenizer = T5Tokenizer.from_pretrained("rinna/japanese-gpt-neox-small")
+tokenizer = AutoTokenizer.from_pretrained("rinna/japanese-gpt-neox-small", use_fast=False)
 model = GPTNeoXForCausalLM.from_pretrained("rinna/japanese-gpt-neox-small")
 ~~~~
 

config.json

@@ -1,42 +1,24 @@
 {
-  "activation_function": "gelu",
   "architectures": [
     "GPTNeoXForCausalLM"
   ],
-  "attn_pdrop": 0.0,
   "bos_token_id": 2,
-  "embd_pdrop": 0.0,
   "eos_token_id": 3,
-  "gpt_j_residual": false,
-  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_size": 768,
   "initializer_range": 0.02,
-  "layer_norm_epsilon": 1e-05,
-  "lm_head_bias": false,
-  "model_type": "gpt-neox",
-  "n_embd": 768,
-  "n_head": 12,
-  "n_inner": null,
-  "n_layer": 12,
-  "n_positions": 2048,
-  "resid_pdrop": 0.0,
-  "rotary": true,
-  "rotary_dim": null,
-  "scale_attn_weights": true,
-  "summary_activation": null,
-  "summary_first_dropout": 0.1,
-  "summary_proj_to_labels": true,
-  "summary_type": "cls_index",
-  "summary_use_proj": true,
-  "task_specific_params": {
-    "text-generation": {
-      "do_sample": true,
-      "max_length": 50,
-      "temperature": 1.0
-    }
-  },
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 2048,
+  "model_type": "gpt_neox",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "rotary_emb_base": 10000,
+  "rotary_pct": 1.0,
   "tie_word_embeddings": false,
   "tokenizer_class": "T5Tokenizer",
   "torch_dtype": "float32",
   "use_cache": true,
+  "use_parallel_residual": false,
   "vocab_size": 44416
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b59cd532d22f3dad2a3afea49ba4eb66c084ca13e3e703bf3090a558ac9a1ced
-size 663506729
+oid sha256:58e3f25be62e0e5e860b36a4e9b1ef0bd9299378a488c1dafb48526222076d5b
+size 663506641