anrilombard
commited on
Commit
•
d3721c7
1
Parent(s):
eb42b4e
Upload config.json with huggingface_hub
Browse files- config.json +37 -0
config.json
ADDED
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"model_type": "mamba",
|
3 |
+
"d_model": 512,
|
4 |
+
"d_intermediate": 0,
|
5 |
+
"n_layer": 6,
|
6 |
+
"vocab_size": 1880,
|
7 |
+
"ssm_cfg": {
|
8 |
+
"layer": "Mamba2"
|
9 |
+
},
|
10 |
+
"attn_layer_idx": [2, 5],
|
11 |
+
"attn_cfg": {
|
12 |
+
"num_heads": 8,
|
13 |
+
"head_dim": 64,
|
14 |
+
"rotary_emb_dim": 32,
|
15 |
+
"causal": true,
|
16 |
+
"softmax_scale": null,
|
17 |
+
"qkv_proj_bias": true,
|
18 |
+
"out_proj_bias": true,
|
19 |
+
"mlp_dim": 0
|
20 |
+
},
|
21 |
+
"rms_norm": true,
|
22 |
+
"residual_in_fp32": true,
|
23 |
+
"fused_add_norm": true,
|
24 |
+
"pad_vocab_size_multiple": 8,
|
25 |
+
"tie_embeddings": true,
|
26 |
+
"dropout_rate": 0.1,
|
27 |
+
"eos_token_id": 2,
|
28 |
+
"bos_token_id": 1,
|
29 |
+
"pad_token_id": 3,
|
30 |
+
"num_labels": 1,
|
31 |
+
"summary_type": "cls_index",
|
32 |
+
"summary_use_proj": true,
|
33 |
+
"summary_activation": "tanh",
|
34 |
+
"summary_proj_to_labels": true,
|
35 |
+
"summary_first_dropout": 0.1,
|
36 |
+
"summary_hidden_size": 128
|
37 |
+
}
|