Add files using upload-large-folder tool

config.json

@@ -12,65 +12,19 @@
   "freeze_speech_adapter": false,
   "hidden_act": "silu",
   "hidden_size": 3584,
-  "image_aspect_ratio": "anyres_genli",
-  "image_crop_resolution": 224,
-  "image_grid_pinpoints": [
-    [
-      336,
-      672
-    ],
-    [
-      672,
-      336
-    ],
-    [
-      672,
-      672
-    ],
-    [
-      1008,
-      336
-    ],
-    [
-      336,
-      1008
-    ],
-    [
-      1008,
-      672
-    ],
-    [
-      672,
-      1008
-    ],
-    [
-      1008,
-      1008
-    ]
-  ],
-  "image_split_resolution": 224,
   "initializer_range": 0.02,
   "intermediate_size": 18944,
   "max_position_embeddings": 32768,
   "max_window_layers": 28,
   "mm_hidden_size": 1152,
-  "mm_newline_position": "no_token",
-  "mm_patch_merge_type": "genli",
-  "mm_pooling_position": "before",
-  "mm_projector_lr": null,
   "mm_projector_type": "ola_mlp",
-  "mm_resampler_type": null,
-  "mm_spatial_pool_mode": "average",
-  "mm_spatial_pool_out_channels": null,
-  "mm_spatial_pool_stride": 2,
   "mm_use_im_patch_token": false,
   "mm_use_im_start_end": false,
   "mm_vision_select_feature": "patch",
   "mm_vision_select_layer": -1,
-  "music_encoder": "/path/to/beats",
-  "mm_vision_tower": "/path/to/oryx_vit",
+  "music_encoder": "./pretrained/BEATs_iter3_plus_AS2M_finetuned_on_AS2M_cpt2.pt",
+  "mm_vision_tower": "./pretrained/oryx_vit.pth",
   "mm_vision_tower_lr": null,
-  "mm_vision_tower_two": null,
   "modality_max_length": "None",
   "model_type": "ola_qwen",
   "num_attention_heads": 28,
@@ -80,7 +34,7 @@
   "rope_scaling": null,
   "rope_theta": 1000000.0,
   "sliding_window": null,
-  "speech_encoder": "/path/to/whisper",
+  "speech_encoder": "./pretrained/large-v3.pt",
   "speech_encoder_ds_rate": 10,
   "speech_encoder_hidden_size": 2048,
   "speech_encoder_type": "dual",