(Trained with Unsloth)
Browse files- config.json +7 -7
config.json
CHANGED
@@ -9,18 +9,18 @@
|
|
9 |
"cache_implementation": "hybrid",
|
10 |
"eos_token_id": 1,
|
11 |
"final_logit_softcapping": null,
|
12 |
-
"head_dim":
|
13 |
"hidden_activation": "gelu_pytorch_tanh",
|
14 |
-
"hidden_size":
|
15 |
"initializer_range": 0.02,
|
16 |
-
"intermediate_size":
|
17 |
"max_position_embeddings": 131072,
|
18 |
"model_type": "gemma3_text",
|
19 |
-
"num_attention_heads":
|
20 |
-
"num_hidden_layers":
|
21 |
-
"num_key_value_heads":
|
22 |
"pad_token_id": 0,
|
23 |
-
"query_pre_attn_scalar":
|
24 |
"rms_norm_eps": 1e-06,
|
25 |
"rope_local_base_freq": 10000.0,
|
26 |
"rope_scaling": {
|
|
|
9 |
"cache_implementation": "hybrid",
|
10 |
"eos_token_id": 1,
|
11 |
"final_logit_softcapping": null,
|
12 |
+
"head_dim": 128,
|
13 |
"hidden_activation": "gelu_pytorch_tanh",
|
14 |
+
"hidden_size": 5376,
|
15 |
"initializer_range": 0.02,
|
16 |
+
"intermediate_size": 21504,
|
17 |
"max_position_embeddings": 131072,
|
18 |
"model_type": "gemma3_text",
|
19 |
+
"num_attention_heads": 32,
|
20 |
+
"num_hidden_layers": 62,
|
21 |
+
"num_key_value_heads": 16,
|
22 |
"pad_token_id": 0,
|
23 |
+
"query_pre_attn_scalar": 168,
|
24 |
"rms_norm_eps": 1e-06,
|
25 |
"rope_local_base_freq": 10000.0,
|
26 |
"rope_scaling": {
|