| { |
| "module": "keras_hub.src.models.qwen.qwen_causal_lm", |
| "class_name": "QwenCausalLM", |
| "config": { |
| "backbone": { |
| "module": "keras_hub.src.models.qwen.qwen_backbone", |
| "class_name": "QwenBackbone", |
| "config": { |
| "name": "qwen_backbone", |
| "trainable": true, |
| "dtype": { |
| "module": "keras", |
| "class_name": "DTypePolicy", |
| "config": { |
| "name": "float32" |
| }, |
| "registered_name": null |
| }, |
| "vocabulary_size": 152064, |
| "num_layers": 28, |
| "num_query_heads": 28, |
| "hidden_dim": 3584, |
| "intermediate_dim": 18944, |
| "rope_max_wavelength": 1000000.0, |
| "rope_scaling_factor": 1.0, |
| "num_key_value_heads": 4, |
| "layer_norm_epsilon": 1e-06, |
| "dropout": 0, |
| "tie_word_embeddings": false, |
| "use_sliding_window_attention": false, |
| "sliding_window_size": 131072 |
| }, |
| "registered_name": "keras_hub>QwenBackbone" |
| }, |
| "preprocessor": { |
| "module": "keras_hub.src.models.qwen.qwen_causal_lm_preprocessor", |
| "class_name": "QwenCausalLMPreprocessor", |
| "config": { |
| "name": "qwen_causal_lm_preprocessor_2", |
| "trainable": true, |
| "dtype": { |
| "module": "keras", |
| "class_name": "DTypePolicy", |
| "config": { |
| "name": "float32" |
| }, |
| "registered_name": null |
| }, |
| "tokenizer": { |
| "module": "keras_hub.src.models.qwen.qwen_tokenizer", |
| "class_name": "QwenTokenizer", |
| "config": { |
| "name": "qwen_tokenizer", |
| "trainable": true, |
| "dtype": { |
| "module": "keras", |
| "class_name": "DTypePolicy", |
| "config": { |
| "name": "int32" |
| }, |
| "registered_name": null |
| }, |
| "config_file": "tokenizer.json", |
| "sequence_length": null, |
| "add_prefix_space": false, |
| "unsplittable_tokens": [ |
| "<|file_sep|>", |
| "<|vision_start|>", |
| "<|quad_start|>", |
| "<|object_ref_start|>", |
| "<|fim_pad|>", |
| "<|fim_suffix|>", |
| "<|image_pad|>", |
| "<|fim_middle|>", |
| "</tool_call>", |
| "<|box_end|>", |
| "<|object_ref_end|>", |
| "<|quad_end|>", |
| "<|vision_pad|>", |
| "<|repo_name|>", |
| "<|box_start|>", |
| "<|fim_prefix|>", |
| "<|vision_end|>", |
| "<tool_call>", |
| "<|im_end|>", |
| "<|endoftext|>", |
| "<|im_start|>", |
| "<|video_pad|>" |
| ] |
| }, |
| "registered_name": "keras_hub>QwenTokenizer" |
| }, |
| "config_file": "preprocessor.json", |
| "sequence_length": 1024, |
| "add_start_token": true, |
| "add_end_token": true |
| }, |
| "registered_name": "keras_hub>QwenCausalLMPreprocessor" |
| }, |
| "name": "qwen_causal_lm" |
| }, |
| "registered_name": "keras_hub>QwenCausalLM" |
| } |