gnukeith/Ocelot-MLX
08
1{2 "architectures": [3 "Qwen3VLForConditionalGeneration"4 ],5 "do_sample": true,6 "dtype": "bfloat16",7 "eos_token_id": [8 151645,9 15164310 ],11 "generation_config": {12 "bos_token_id": 151643,13 "do_sample": true,14 "eos_token_id": [15 151645,16 15164317 ],18 "pad_token_id": 151643,19 "temperature": 0.7,20 "top_k": 20,21 "top_p": 0.8,22 "transformers_version": "4.57.6"23 },24 "image_token_id": 151655,25 "model_type": "qwen3_vl",26 "quantization": {27 "group_size": 64,28 "bits": 4,29 "mode": "affine"30 },31 "quantization_config": {32 "group_size": 64,33 "bits": 4,34 "mode": "affine"35 },36 "temperature": 0.7,37 "text_config": {38 "attention_bias": false,39 "attention_dropout": 0.0,40 "bos_token_id": 151643,41 "dtype": "bfloat16",42 "eos_token_id": 151645,43 "head_dim": 128,44 "hidden_act": "silu",45 "hidden_size": 2560,46 "initializer_range": 0.02,47 "intermediate_size": 9728,48 "max_position_embeddings": 262144,49 "model_type": "qwen3_vl_text",50 "num_attention_heads": 32,51 "num_hidden_layers": 36,52 "num_key_value_heads": 8,53 "rms_norm_eps": 1e-06,54 "rope_scaling": {55 "mrope_interleaved": true,56 "mrope_section": [57 24,58 20,59 2060 ],61 "rope_type": "default"62 },63 "rope_theta": 5000000,64 "tie_word_embeddings": true,65 "use_cache": true,66 "vocab_size": 15193667 },68 "tie_word_embeddings": true,69 "top_k": 20,70 "top_p": 0.8,71 "transformers_version": "4.57.6",72 "video_token_id": 151656,73 "vision_config": {74 "deepstack_visual_indexes": [75 5,76 11,77 1778 ],79 "depth": 24,80 "dtype": "bfloat16",81 "hidden_act": "gelu_pytorch_tanh",82 "hidden_size": 1024,83 "in_channels": 3,84 "initializer_range": 0.02,85 "intermediate_size": 4096,86 "model_type": "qwen3_vl",87 "num_heads": 16,88 "num_position_embeddings": 2304,89 "out_hidden_size": 2560,90 "patch_size": 16,91 "spatial_merge_size": 2,92 "temporal_patch_size": 293 },94 "vision_end_token_id": 151653,95 "vision_start_token_id": 15165296}