Upload FlaxCohereForCausalLM
Browse files- .gitattributes +1 -0
- config.json +83 -0
- easydel-model.parameters +3 -0
- generation_config.json +7 -0
.gitattributes
CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
36 |
+
easydel-model.parameters filter=lfs diff=lfs merge=lfs -text
|
config.json
ADDED
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"architectures": [
|
3 |
+
"CohereForCausalLM"
|
4 |
+
],
|
5 |
+
"attention_axis_name": "sp",
|
6 |
+
"attention_bias": false,
|
7 |
+
"attention_dropout": 0.0,
|
8 |
+
"attn_mechanism": "sharded_vanilla",
|
9 |
+
"axis_dims": [
|
10 |
+
1,
|
11 |
+
-1,
|
12 |
+
1,
|
13 |
+
1
|
14 |
+
],
|
15 |
+
"axis_names": [
|
16 |
+
"dp",
|
17 |
+
"fsdp",
|
18 |
+
"tp",
|
19 |
+
"sp"
|
20 |
+
],
|
21 |
+
"backend": null,
|
22 |
+
"bits": null,
|
23 |
+
"block_b": 1,
|
24 |
+
"block_k": 128,
|
25 |
+
"block_k_dkv": 128,
|
26 |
+
"block_k_dq": 128,
|
27 |
+
"block_k_major": 128,
|
28 |
+
"block_k_major_dkv": 128,
|
29 |
+
"block_k_major_dq": 128,
|
30 |
+
"block_q": 128,
|
31 |
+
"block_q_dkv": 128,
|
32 |
+
"block_q_dq": 128,
|
33 |
+
"block_q_major_dkv": 128,
|
34 |
+
"bos_token_id": 5,
|
35 |
+
"easy_method": "train",
|
36 |
+
"eos_token_id": 255001,
|
37 |
+
"flash_attention_backward_pass_impl": "triton",
|
38 |
+
"gradient_checkpointing": "nothing_saveable",
|
39 |
+
"hidden_act": "silu",
|
40 |
+
"hidden_size": 4096,
|
41 |
+
"initializer_range": 0.02,
|
42 |
+
"intermediate_size": 14336,
|
43 |
+
"layer_norm_eps": 1e-05,
|
44 |
+
"logit_scale": 0.0625,
|
45 |
+
"max_position_embeddings": 8192,
|
46 |
+
"model_type": "cohere",
|
47 |
+
"num_attention_heads": 32,
|
48 |
+
"num_hidden_layers": 32,
|
49 |
+
"num_key_value_heads": 8,
|
50 |
+
"pad_token_id": 0,
|
51 |
+
"partition_axis": [
|
52 |
+
[
|
53 |
+
"fsdp",
|
54 |
+
"dp"
|
55 |
+
],
|
56 |
+
"sp",
|
57 |
+
"sp",
|
58 |
+
"tp",
|
59 |
+
"sp",
|
60 |
+
"tp",
|
61 |
+
null,
|
62 |
+
null,
|
63 |
+
null,
|
64 |
+
null,
|
65 |
+
"tp",
|
66 |
+
"sp",
|
67 |
+
null
|
68 |
+
],
|
69 |
+
"quantize_kv_cache": false,
|
70 |
+
"rope_theta": 10000,
|
71 |
+
"scan_attention_layers": false,
|
72 |
+
"scan_mlp_chunk_size": 1024,
|
73 |
+
"scan_ring_attention": true,
|
74 |
+
"shard_attention_computation": true,
|
75 |
+
"torch_dtype": "float16",
|
76 |
+
"transformers_version": "4.41.2",
|
77 |
+
"use_cache": true,
|
78 |
+
"use_qk_norm": false,
|
79 |
+
"use_scan_mlp": false,
|
80 |
+
"use_sharded_kv_caching": true,
|
81 |
+
"use_sharding_constraint": false,
|
82 |
+
"vocab_size": 256000
|
83 |
+
}
|
easydel-model.parameters
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4cdb92ba99eb14863dd505196b7d32f8b99f50b34b9a2c47ffa2471f24f64594
|
3 |
+
size 16056096152
|
generation_config.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"bos_token_id": 5,
|
4 |
+
"eos_token_id": 255001,
|
5 |
+
"pad_token_id": 0,
|
6 |
+
"transformers_version": "4.41.2"
|
7 |
+
}
|