AdamF92 commited on
Commit
34cccbc
·
verified ·
1 Parent(s): dfa06cb

In progress training - batch: 4096

Browse files
Files changed (2) hide show
  1. config.json +43 -0
  2. model.safetensors +3 -0
config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "att_groups": 4,
3
+ "att_heads": 16,
4
+ "att_query_groups": 8,
5
+ "cross_att_type": "sqa",
6
+ "dense_layer_dim": 1024,
7
+ "embed_dim": 256,
8
+ "ff_activation": "silu",
9
+ "ff_dim": 96,
10
+ "ff_dropout": 0.0,
11
+ "final_stateless_layers_config": [
12
+ "moe"
13
+ ],
14
+ "head_norm_type": "rms_norm",
15
+ "moe_bias_mode": "global",
16
+ "moe_grouped_gemm": true,
17
+ "moe_shared_experts_bias_mode": "global",
18
+ "moe_top_k": 10,
19
+ "moe_use_cutlass_grouped_gemm": true,
20
+ "moe_use_weighted_shared_experts": false,
21
+ "num_experts": 384,
22
+ "num_layers": 16,
23
+ "num_shared_experts": 2,
24
+ "rope_base": 100000,
25
+ "router_amp": true,
26
+ "self_att_type": "sqa",
27
+ "seq_len": 8192,
28
+ "shared_expert_dim": 192,
29
+ "stateless_layers_config": [
30
+ "dense",
31
+ "moe"
32
+ ],
33
+ "stm_size": 4096,
34
+ "use_attention_output_bias": false,
35
+ "use_flash_attention": true,
36
+ "use_gated": true,
37
+ "use_gated_attention": true,
38
+ "use_gated_cross_attention": false,
39
+ "use_head_norm": true,
40
+ "use_moe": true,
41
+ "use_vectorized_moe": true,
42
+ "vocab_size": 65536
43
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:af61e01526b02a009486f5b69025f14b4988c718031fa6b505bfae29f4170e30
3
+ size 1144123896