Aissa Bakhil commited on
Commit ·
d5de630
1
Parent(s): 8e748e1
overwrite with latest pruned model
Browse files- config.json +8 -8
- model.safetensors +3 -0
config.json
CHANGED
|
@@ -3,21 +3,21 @@
|
|
| 3 |
"LlamaForCausalLM"
|
| 4 |
],
|
| 5 |
"attention_bias": false,
|
| 6 |
-
"attention_dropout": 0.
|
| 7 |
"bos_token_id": 0,
|
| 8 |
"eos_token_id": 0,
|
| 9 |
-
"head_dim":
|
| 10 |
"hidden_act": "silu",
|
| 11 |
-
"hidden_size":
|
| 12 |
-
"initializer_range": 0.
|
| 13 |
-
"intermediate_size":
|
| 14 |
"is_llama_config": true,
|
| 15 |
"max_position_embeddings": 8192,
|
| 16 |
"mlp_bias": false,
|
| 17 |
"model_type": "llama",
|
| 18 |
-
"num_attention_heads":
|
| 19 |
-
"num_hidden_layers":
|
| 20 |
-
"num_key_value_heads":
|
| 21 |
"pretraining_tp": 1,
|
| 22 |
"rms_norm_eps": 1e-05,
|
| 23 |
"rope_interleaved": false,
|
|
|
|
| 3 |
"LlamaForCausalLM"
|
| 4 |
],
|
| 5 |
"attention_bias": false,
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
"bos_token_id": 0,
|
| 8 |
"eos_token_id": 0,
|
| 9 |
+
"head_dim": 64,
|
| 10 |
"hidden_act": "silu",
|
| 11 |
+
"hidden_size": 576,
|
| 12 |
+
"initializer_range": 0.041666666666666664,
|
| 13 |
+
"intermediate_size": 1229,
|
| 14 |
"is_llama_config": true,
|
| 15 |
"max_position_embeddings": 8192,
|
| 16 |
"mlp_bias": false,
|
| 17 |
"model_type": "llama",
|
| 18 |
+
"num_attention_heads": 9,
|
| 19 |
+
"num_hidden_layers": 23,
|
| 20 |
+
"num_key_value_heads": 3,
|
| 21 |
"pretraining_tp": 1,
|
| 22 |
"rms_norm_eps": 1e-05,
|
| 23 |
"rope_interleaved": false,
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d28a1dde5d4ebd426701f791148793d77b5cc8d5a04b8fbfbd3d2ed61102eb15
|
| 3 |
+
size 390155032
|