Update config.json

config.json

@@ -1,23 +1,25 @@
 {
-  "_name_or_path": "microsoft/llava-med-v1.5-mistral-7b",
+  "_name_or_path": "/scratch/nshaheen/project/main_models/llava_no_tags_8epochs",
   "architectures": [
     "LlavaLlamaForCausalLM"
   ],
+  "attention_bias": false,
   "attention_dropout": 0.0,
-  "bos_token_id": 1,
-  "eos_token_id": 2,
-  "feature_outs": "encoder+decoder",
+  "bos_token_id": 0,
+  "eos_token_id": 1,
   "freeze_mm_mlp_adapter": false,
   "hidden_act": "silu",
   "hidden_size": 4096,
   "image_aspect_ratio": "pad",
-  "img_size": 640,
   "initializer_range": 0.02,
-  "intermediate_size": 14336,
-  "max_position_embeddings": 32768,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 2048,
+  "max_sequence_length": 2048,
+  "mlp_bias": false,
   "mm_hidden_size": 1024,
-  "mm_projector_lr": null,
-  "mm_projector_type": "mlp2x_gelu",
+  "mm_patch_merge_type": "flat",
+  "mm_projector_lr": 2e-05,
+  "mm_projector_type": "linear",
   "mm_use_im_patch_token": false,
   "mm_use_im_start_end": false,
   "mm_vision_select_feature": "patch",
@@ -26,23 +28,20 @@
   "model_type": "llava_llama",
   "num_attention_heads": 32,
   "num_hidden_layers": 32,
-  "num_key_value_heads": 8,
-  "proj_vis_to_txt_tokens": false,
-  "prompt_segtok_w_instruct": false,
-  "rms_norm_eps": 1e-05,
-  "rope_theta": 1000000.0,
-  "segtok_posembed": "sincos",
-  "sliding_window": null,
+  "num_key_value_heads": 32,
+  "pad_token_id": -1,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
   "tie_word_embeddings": false,
   "tokenizer_model_max_length": 2048,
   "tokenizer_padding_side": "right",
   "torch_dtype": "float16",
-  "transformers_version": "4.39.3",
+  "transformers_version": "4.41.2",
   "tune_mm_mlp_adapter": false,
-  "tune_vision_tokenizer": "none",
   "use_cache": true,
   "use_mm_proj": true,
-  "vision_backbone": "convnextlarge",
-  "vision_tokenizer_lr": null,
-  "vocab_size": 32000
+  "vocab_size": 32004
 }