Update config.json
Browse files- config.json +7 -10
config.json
CHANGED
|
@@ -16,13 +16,10 @@
|
|
| 16 |
"head_dim": 128,
|
| 17 |
"max_position_embeddings": 131072,
|
| 18 |
"rope_scaling": {
|
| 19 |
-
"
|
| 20 |
-
"
|
| 21 |
-
"low_freq_factor": 1.0,
|
| 22 |
-
"original_max_position_embeddings": 8192,
|
| 23 |
-
"rope_type": "llama3"
|
| 24 |
},
|
| 25 |
-
|
| 26 |
"hidden_act": "silu",
|
| 27 |
"initializer_range": 0.02,
|
| 28 |
"rms_norm_eps": 1e-05,
|
|
@@ -34,16 +31,16 @@
|
|
| 34 |
"use_cache": true,
|
| 35 |
"shared_input_output_embedding": true,
|
| 36 |
"attention_dropout": 0.0,
|
| 37 |
-
|
| 38 |
"attn_dropout_prob": 0.1,
|
| 39 |
-
|
| 40 |
"ffn_dropout_prob": 0.1,
|
| 41 |
"output_projection": true,
|
| 42 |
"position_embedding_type": "rotary",
|
| 43 |
"use_rope": true,
|
| 44 |
-
|
| 45 |
"multi_query_attention": false,
|
| 46 |
"output_attentions": false,
|
| 47 |
"output_hidden_states": false,
|
| 48 |
"layerdrop": 0.0
|
| 49 |
-
}
|
|
|
|
| 16 |
"head_dim": 128,
|
| 17 |
"max_position_embeddings": 131072,
|
| 18 |
"rope_scaling": {
|
| 19 |
+
"type": "linear",
|
| 20 |
+
"factor": 32.0
|
|
|
|
|
|
|
|
|
|
| 21 |
},
|
| 22 |
+
"rope_theta": 500000.0,
|
| 23 |
"hidden_act": "silu",
|
| 24 |
"initializer_range": 0.02,
|
| 25 |
"rms_norm_eps": 1e-05,
|
|
|
|
| 31 |
"use_cache": true,
|
| 32 |
"shared_input_output_embedding": true,
|
| 33 |
"attention_dropout": 0.0,
|
| 34 |
+
"hidden_dropout_prob": 0.1,
|
| 35 |
"attn_dropout_prob": 0.1,
|
| 36 |
+
"residual_dropout_prob": 0.0,
|
| 37 |
"ffn_dropout_prob": 0.1,
|
| 38 |
"output_projection": true,
|
| 39 |
"position_embedding_type": "rotary",
|
| 40 |
"use_rope": true,
|
| 41 |
+
"gradient_checkpointing": false,
|
| 42 |
"multi_query_attention": false,
|
| 43 |
"output_attentions": false,
|
| 44 |
"output_hidden_states": false,
|
| 45 |
"layerdrop": 0.0
|
| 46 |
+
}
|