Update model files
Browse files
checkpoint-2700/config.json
CHANGED
|
@@ -2,6 +2,11 @@
|
|
| 2 |
"architectures": [
|
| 3 |
"SharedSpaceDecoderForCausalLM"
|
| 4 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
"attention_backend": "flash_attention_2",
|
| 6 |
"attention_bias": false,
|
| 7 |
"attention_dropout_prob": 0.1,
|
|
@@ -18,7 +23,6 @@
|
|
| 18 |
"kv_shared_dim": null,
|
| 19 |
"layer_norm_eps": 1e-12,
|
| 20 |
"max_position_embeddings": 1024,
|
| 21 |
-
"model_type": "shared_subspace_decoder",
|
| 22 |
"nope_dims": 32,
|
| 23 |
"norm_type": "rmsnorm",
|
| 24 |
"num_attention_heads": 12,
|
|
@@ -40,4 +44,4 @@
|
|
| 40 |
"vocab_rank": null,
|
| 41 |
"vocab_size": 50257,
|
| 42 |
"vocab_subspace": false
|
| 43 |
-
}
|
|
|
|
| 2 |
"architectures": [
|
| 3 |
"SharedSpaceDecoderForCausalLM"
|
| 4 |
],
|
| 5 |
+
"model_type": "shared_subspace_decoder",
|
| 6 |
+
"auto_map": {
|
| 7 |
+
"AutoConfig": "shared_space_config.SharedSpaceDecoderConfig",
|
| 8 |
+
"AutoModelForCausalLM": "task_heads.SharedSpaceDecoderForCausalLM"
|
| 9 |
+
},
|
| 10 |
"attention_backend": "flash_attention_2",
|
| 11 |
"attention_bias": false,
|
| 12 |
"attention_dropout_prob": 0.1,
|
|
|
|
| 23 |
"kv_shared_dim": null,
|
| 24 |
"layer_norm_eps": 1e-12,
|
| 25 |
"max_position_embeddings": 1024,
|
|
|
|
| 26 |
"nope_dims": 32,
|
| 27 |
"norm_type": "rmsnorm",
|
| 28 |
"num_attention_heads": 12,
|
|
|
|
| 44 |
"vocab_rank": null,
|
| 45 |
"vocab_size": 50257,
|
| 46 |
"vocab_subspace": false
|
| 47 |
+
}
|
checkpoint-3000/config.json
CHANGED
|
@@ -2,6 +2,11 @@
|
|
| 2 |
"architectures": [
|
| 3 |
"SharedSpaceDecoderForCausalLM"
|
| 4 |
],
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
"attention_backend": "flash_attention_2",
|
| 6 |
"attention_bias": false,
|
| 7 |
"attention_dropout_prob": 0.1,
|
|
@@ -18,7 +23,6 @@
|
|
| 18 |
"kv_shared_dim": null,
|
| 19 |
"layer_norm_eps": 1e-12,
|
| 20 |
"max_position_embeddings": 1024,
|
| 21 |
-
"model_type": "shared_subspace_decoder",
|
| 22 |
"nope_dims": 32,
|
| 23 |
"norm_type": "rmsnorm",
|
| 24 |
"num_attention_heads": 12,
|
|
@@ -40,4 +44,4 @@
|
|
| 40 |
"vocab_rank": null,
|
| 41 |
"vocab_size": 50257,
|
| 42 |
"vocab_subspace": false
|
| 43 |
-
}
|
|
|
|
| 2 |
"architectures": [
|
| 3 |
"SharedSpaceDecoderForCausalLM"
|
| 4 |
],
|
| 5 |
+
"model_type": "shared_subspace_decoder",
|
| 6 |
+
"auto_map": {
|
| 7 |
+
"AutoConfig": "shared_space_config.SharedSpaceDecoderConfig",
|
| 8 |
+
"AutoModelForCausalLM": "task_heads.SharedSpaceDecoderForCausalLM"
|
| 9 |
+
},
|
| 10 |
"attention_backend": "flash_attention_2",
|
| 11 |
"attention_bias": false,
|
| 12 |
"attention_dropout_prob": 0.1,
|
|
|
|
| 23 |
"kv_shared_dim": null,
|
| 24 |
"layer_norm_eps": 1e-12,
|
| 25 |
"max_position_embeddings": 1024,
|
|
|
|
| 26 |
"nope_dims": 32,
|
| 27 |
"norm_type": "rmsnorm",
|
| 28 |
"num_attention_heads": 12,
|
|
|
|
| 44 |
"vocab_rank": null,
|
| 45 |
"vocab_size": 50257,
|
| 46 |
"vocab_subspace": false
|
| 47 |
+
}
|