| { | |
| "architectures": [ | |
| "Qwen2ForCausalLM" | |
| ], | |
| "attention_dropout": 0.0, | |
| "max_position_embeddings": 32768, | |
| "max_window_layers": 35, | |
| "model_type": "qwen2", | |
| "num_attention_heads": 40, | |
| "num_hidden_layers": 40, | |
| "num_key_value_heads": 40, | |
| "number_rep_kv": 1, | |
| "resid_pdrop": 0.0, | |
| "rms_norm_eps": 1e-06, | |
| "rope_scaling": null, | |
| "rope_theta": 10000.0, | |
| "scan_attention_layers": false, | |
| "scan_layers": true, | |
| "scan_mlp_chunk_size": 1024, | |
| "scan_ring_attention": false, | |
| "shard_attention_computation": true, | |
| "sliding_window": 32768, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.41.0.dev0", | |
| "use_cache": true, | |
| "use_scan_mlp": false, | |
| "use_sharded_kv_caching": true, | |
| "use_sharding_constraint": false, | |
| "use_sliding_window": false, | |
| "vocab_size": 152064 | |
| } | |