Update config.json
Browse files- config.json +3 -2
config.json
CHANGED
|
@@ -20,6 +20,7 @@
|
|
| 20 |
"kv_lora_rank": 512,
|
| 21 |
"max_position_embeddings": 163840,
|
| 22 |
"model_type": "deepseek_v2",
|
|
|
|
| 23 |
"moe_intermediate_size": 1408,
|
| 24 |
"moe_layer_freq": 1,
|
| 25 |
"n_group": 1,
|
|
@@ -27,7 +28,7 @@
|
|
| 27 |
"n_shared_experts": 2,
|
| 28 |
"norm_topk_prob": false,
|
| 29 |
"num_attention_heads": 16,
|
| 30 |
-
"num_experts_per_tok":
|
| 31 |
"num_hidden_layers": 27,
|
| 32 |
"num_key_value_heads": 16,
|
| 33 |
"pretraining_tp": 1,
|
|
@@ -45,7 +46,7 @@
|
|
| 45 |
"type": "yarn"
|
| 46 |
},
|
| 47 |
"rope_theta": 10000,
|
| 48 |
-
"routed_scaling_factor": 1.
|
| 49 |
"scoring_func": "softmax",
|
| 50 |
"seq_aux": true,
|
| 51 |
"tie_word_embeddings": false,
|
|
|
|
| 20 |
"kv_lora_rank": 512,
|
| 21 |
"max_position_embeddings": 163840,
|
| 22 |
"model_type": "deepseek_v2",
|
| 23 |
+
"_name_or_path": "SarvaCode-16B-Indigenous",
|
| 24 |
"moe_intermediate_size": 1408,
|
| 25 |
"moe_layer_freq": 1,
|
| 26 |
"n_group": 1,
|
|
|
|
| 28 |
"n_shared_experts": 2,
|
| 29 |
"norm_topk_prob": false,
|
| 30 |
"num_attention_heads": 16,
|
| 31 |
+
"num_experts_per_tok": 8,
|
| 32 |
"num_hidden_layers": 27,
|
| 33 |
"num_key_value_heads": 16,
|
| 34 |
"pretraining_tp": 1,
|
|
|
|
| 46 |
"type": "yarn"
|
| 47 |
},
|
| 48 |
"rope_theta": 10000,
|
| 49 |
+
"routed_scaling_factor": 1.5,
|
| 50 |
"scoring_func": "softmax",
|
| 51 |
"seq_aux": true,
|
| 52 |
"tie_word_embeddings": false,
|