Update config.json
Browse files- config.json +2 -2
config.json
CHANGED
@@ -10,7 +10,7 @@
|
|
10 |
"hidden_size": 2048,
|
11 |
"initializer_range": 0.02,
|
12 |
"intermediate_size": 1024,
|
13 |
-
"max_position_embeddings":
|
14 |
"model_type": "olmoe",
|
15 |
"norm_topk_prob": false,
|
16 |
"num_attention_heads": 16,
|
@@ -22,7 +22,7 @@
|
|
22 |
"pad_token_id": 1,
|
23 |
"rope_scaling": null,
|
24 |
"rope_theta": 10000.0,
|
25 |
-
"router_aux_loss_coef": 0.
|
26 |
"tie_word_embeddings": false,
|
27 |
"torch_dtype": "bfloat16",
|
28 |
"transformers_version": "4.43.0.dev0",
|
|
|
10 |
"hidden_size": 2048,
|
11 |
"initializer_range": 0.02,
|
12 |
"intermediate_size": 1024,
|
13 |
+
"max_position_embeddings": 16386,
|
14 |
"model_type": "olmoe",
|
15 |
"norm_topk_prob": false,
|
16 |
"num_attention_heads": 16,
|
|
|
22 |
"pad_token_id": 1,
|
23 |
"rope_scaling": null,
|
24 |
"rope_theta": 10000.0,
|
25 |
+
"router_aux_loss_coef": 0.0,
|
26 |
"tie_word_embeddings": false,
|
27 |
"torch_dtype": "bfloat16",
|
28 |
"transformers_version": "4.43.0.dev0",
|