Update config.json
Browse files- config.json +2 -21
config.json
CHANGED
|
@@ -9,9 +9,6 @@
|
|
| 9 |
"projector_type": "mlp_gelu"
|
| 10 |
}
|
| 11 |
},
|
| 12 |
-
"architectures": [
|
| 13 |
-
"MultiModalityCausalLM"
|
| 14 |
-
],
|
| 15 |
"gen_aligner_config": {
|
| 16 |
"cls": "MlpProjector",
|
| 17 |
"model_type": "gen_aligner",
|
|
@@ -40,31 +37,15 @@
|
|
| 40 |
}
|
| 41 |
},
|
| 42 |
"language_config": {
|
| 43 |
-
"_attn_implementation_autoset": true,
|
| 44 |
-
"attention_bias": false,
|
| 45 |
-
"attention_dropout": 0.0,
|
| 46 |
-
"head_dim": 128,
|
| 47 |
-
"hidden_act": "silu",
|
| 48 |
-
"hidden_size": 4096,
|
| 49 |
-
"initializer_range": 0.02,
|
| 50 |
-
"intermediate_size": 11008,
|
| 51 |
"max_position_embeddings": 16384,
|
| 52 |
-
"mlp_bias": false,
|
| 53 |
"model_type": "llama",
|
| 54 |
-
"num_attention_heads": 32,
|
| 55 |
"num_hidden_layers": 30,
|
| 56 |
-
"num_key_value_heads": 32,
|
| 57 |
-
"pretraining_tp": 1,
|
| 58 |
-
"rms_norm_eps": 1e-06,
|
| 59 |
-
"rope_scaling": null,
|
| 60 |
-
"rope_theta": 10000.0,
|
| 61 |
"torch_dtype": "bfloat16",
|
| 62 |
-
"use_cache": true,
|
| 63 |
"vocab_size": 102400
|
| 64 |
},
|
| 65 |
"model_type": "multi_modality",
|
| 66 |
"torch_dtype": "bfloat16",
|
| 67 |
-
"transformers_version": "4.
|
| 68 |
"vision_config": {
|
| 69 |
"cls": "CLIPVisionTower",
|
| 70 |
"model_type": "vision",
|
|
@@ -75,4 +56,4 @@
|
|
| 75 |
"select_layer": -1
|
| 76 |
}
|
| 77 |
}
|
| 78 |
-
}
|
|
|
|
| 9 |
"projector_type": "mlp_gelu"
|
| 10 |
}
|
| 11 |
},
|
|
|
|
|
|
|
|
|
|
| 12 |
"gen_aligner_config": {
|
| 13 |
"cls": "MlpProjector",
|
| 14 |
"model_type": "gen_aligner",
|
|
|
|
| 37 |
}
|
| 38 |
},
|
| 39 |
"language_config": {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40 |
"max_position_embeddings": 16384,
|
|
|
|
| 41 |
"model_type": "llama",
|
|
|
|
| 42 |
"num_hidden_layers": 30,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 43 |
"torch_dtype": "bfloat16",
|
|
|
|
| 44 |
"vocab_size": 102400
|
| 45 |
},
|
| 46 |
"model_type": "multi_modality",
|
| 47 |
"torch_dtype": "bfloat16",
|
| 48 |
+
"transformers_version": "4.33.1",
|
| 49 |
"vision_config": {
|
| 50 |
"cls": "CLIPVisionTower",
|
| 51 |
"model_type": "vision",
|
|
|
|
| 56 |
"select_layer": -1
|
| 57 |
}
|
| 58 |
}
|
| 59 |
+
}
|