pretrain
Browse files- config-0.json +4 -4
config-0.json
CHANGED
|
@@ -8,15 +8,15 @@
|
|
| 8 |
"eos_token_id": 1,
|
| 9 |
"head_dim": 64,
|
| 10 |
"hidden_act": "silu",
|
| 11 |
-
"hidden_size":
|
| 12 |
"initializer_range": 0.02,
|
| 13 |
-
"intermediate_size":
|
| 14 |
"max_position_embeddings": 131072,
|
| 15 |
"mlp_bias": false,
|
| 16 |
"model_type": "llama",
|
| 17 |
-
"num_attention_heads":
|
| 18 |
"num_hidden_layers": 32,
|
| 19 |
-
"num_key_value_heads":
|
| 20 |
"pretraining_tp": 1,
|
| 21 |
"rms_norm_eps": 1e-05,
|
| 22 |
"rope_scaling": null,
|
|
|
|
| 8 |
"eos_token_id": 1,
|
| 9 |
"head_dim": 64,
|
| 10 |
"hidden_act": "silu",
|
| 11 |
+
"hidden_size": 512,
|
| 12 |
"initializer_range": 0.02,
|
| 13 |
+
"intermediate_size": 1365,
|
| 14 |
"max_position_embeddings": 131072,
|
| 15 |
"mlp_bias": false,
|
| 16 |
"model_type": "llama",
|
| 17 |
+
"num_attention_heads": 8,
|
| 18 |
"num_hidden_layers": 32,
|
| 19 |
+
"num_key_value_heads": 8,
|
| 20 |
"pretraining_tp": 1,
|
| 21 |
"rms_norm_eps": 1e-05,
|
| 22 |
"rope_scaling": null,
|