tarekziade HF Staff
Keep the real checkpoint's weight names, prefix and file format (tensors unchanged; reload verified bit-exact)
89eddb3 verified | { | |
| "fixture": "huggingface-mic/DeepSeek-R1-mic", | |
| "key": "3bd6b377673ade4cffef42b326863c5541aa06225fd97eabc42fda15040f43cb", | |
| "key_fields": { | |
| "architectures": [ | |
| "DeepseekV3ForCausalLM" | |
| ], | |
| "output_hidden_states": false, | |
| "return_dict": true, | |
| "chunk_size_feed_forward": 0, | |
| "is_encoder_decoder": false, | |
| "id2label": { | |
| "0": "LABEL_0", | |
| "1": "LABEL_1" | |
| }, | |
| "label2id": { | |
| "LABEL_0": 0, | |
| "LABEL_1": 1 | |
| }, | |
| "problem_type": null, | |
| "vocab_size": 129280, | |
| "hidden_size": 64, | |
| "intermediate_size": 128, | |
| "moe_intermediate_size": 32, | |
| "num_hidden_layers": 2, | |
| "num_attention_heads": 4, | |
| "num_key_value_heads": 4, | |
| "n_shared_experts": 1, | |
| "n_routed_experts": 4, | |
| "output_router_logits": false, | |
| "routed_scaling_factor": 2.5, | |
| "kv_lora_rank": 16, | |
| "q_lora_rank": 16, | |
| "qk_rope_head_dim": 8, | |
| "v_head_dim": 16, | |
| "qk_nope_head_dim": 16, | |
| "n_group": 1, | |
| "topk_group": 1, | |
| "num_experts_per_tok": 2, | |
| "first_k_dense_replace": 1, | |
| "norm_topk_prob": true, | |
| "hidden_act": "silu", | |
| "max_position_embeddings": 163840, | |
| "initializer_range": 0.02, | |
| "rms_norm_eps": 1e-06, | |
| "use_cache": true, | |
| "pad_token_id": null, | |
| "bos_token_id": 0, | |
| "eos_token_id": 1, | |
| "pretraining_tp": 1, | |
| "tie_word_embeddings": false, | |
| "rope_parameters": { | |
| "beta_fast": 32, | |
| "beta_slow": 1, | |
| "factor": 40, | |
| "mscale": 1.0, | |
| "mscale_all_dim": 1.0, | |
| "original_max_position_embeddings": 4096, | |
| "type": "yarn", | |
| "rope_theta": 10000, | |
| "rope_type": "yarn" | |
| }, | |
| "rope_interleave": true, | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "num_nextn_predict_layers": 1, | |
| "qk_head_dim": 192, | |
| "head_dim": 8, | |
| "auto_map": { | |
| "AutoConfig": "configuration_deepseek.DeepseekV3Config", | |
| "AutoModel": "modeling_deepseek.DeepseekV3Model", | |
| "AutoModelForCausalLM": "modeling_deepseek.DeepseekV3ForCausalLM" | |
| }, | |
| "ep_size": 1, | |
| "model_type": "deepseek_v3", | |
| "moe_layer_freq": 1, | |
| "scoring_func": "sigmoid", | |
| "topk_method": "noaux_tc", | |
| "output_attentions": false, | |
| "dtype": "bfloat16" | |
| }, | |
| "covers": { | |
| "deepseek-ai/DeepSeek-R1": "56d4cbbb4d29f4355bab4b9a39ccb717a14ad5ad" | |
| }, | |
| "built_from": { | |
| "repo": "deepseek-ai/DeepSeek-R1", | |
| "revision": "56d4cbbb4d29f4355bab4b9a39ccb717a14ad5ad" | |
| }, | |
| "seed": 17, | |
| "dtype": "bfloat16", | |
| "builder": "example/build_fixtures.py", | |
| "model_class": "DeepseekV3ForCausalLM", | |
| "parameters": 16624256, | |
| "missing_parameters": [], | |
| "checkpoint_inspection": "parameter headers/indexes; bounded binary fallback", | |
| "numeric_health_scope": "CPU float32/bfloat16/float16; autocast for low-precision losses", | |
| "captured": { | |
| "call": "training step", | |
| "inputs": [ | |
| "attention_mask", | |
| "input_ids", | |
| "labels" | |
| ], | |
| "class": "DeepseekV3ForCausalLM" | |
| }, | |
| "weights": "random", | |
| "gate": { | |
| "random": { | |
| "pass": true, | |
| "accepted_insensitive": {}, | |
| "threshold": 0.001, | |
| "deltas": { | |
| "input_ids:roll_batch": 0.931640625, | |
| "input_ids:reverse": 1.0390625, | |
| "attention_mask:all_ones": null | |
| }, | |
| "weakest": "input_ids:roll_batch", | |
| "no_perturbation_defined": [], | |
| "not_a_forward_parameter": [] | |
| } | |
| }, | |
| "numeric_health": { | |
| "float32": { | |
| "outputs_finite": true, | |
| "loss_finite": true, | |
| "gradients_finite": true, | |
| "finite": true | |
| }, | |
| "bfloat16": { | |
| "outputs_finite": true, | |
| "loss_finite": true, | |
| "gradients_finite": true, | |
| "finite": true | |
| }, | |
| "float16": { | |
| "outputs_finite": true, | |
| "loss_finite": true, | |
| "unscaled_gradients_finite": true, | |
| "finite": true | |
| } | |
| }, | |
| "copied_unchanged": [ | |
| "LICENSE", | |
| "configuration_deepseek.py", | |
| "generation_config.json", | |
| "modeling_deepseek.py", | |
| "tokenizer.json", | |
| "tokenizer_config.json" | |
| ], | |
| "environment": { | |
| "python": "3.12.12", | |
| "machine": "arm64", | |
| "torch": "2.14.1", | |
| "transformers": "5.22.0.dev0", | |
| "tokenizers": "0.23.3" | |
| }, | |
| "sha256": { | |
| "LICENSE": "1c8f573e830ca9b3ebfeb7ace1823146e22b66f99ee223840e7637c9e745e1c7", | |
| "config.json": "38c3aa3e4a1367dae62f73668c736aff5af1eb60abf22bc96ba2ca6717413a55", | |
| "configuration_deepseek.py": "d5126f7a82eaa5fe10c294af7a8f02d59e7145c2dc8837b83ae338c1b72001d2", | |
| "generation_config.json": "3f21d5d199c7ea6b0f807e3d9865a8a9957b76a604c946bce6e73fd0629f8ca3", | |
| "model.safetensors": "3f7c061f15ab96c535aeae066495eb0f098c798c87e714a74a0d06a2db42423b", | |
| "modeling_deepseek.py": "eb6d535742061cf9da7a4232ecfe2f089d5ce20080afc5d2181e9d70ec4af8a4", | |
| "tokenizer.json": "ecb6f9fc369894346f0511f4074ca75cee5cd5f3b06d02f1ba35fcd39f8e121d", | |
| "tokenizer_config.json": "99e80c96269ca4430f1e67f67b5766cb096cbac96fed7a7c611c6ee088b3a039" | |
| }, | |
| "serialization": { | |
| "file": "model.safetensors", | |
| "reverted_conversions": [ | |
| "WeightConverter: ['mlp.experts.*.gate_proj.weight', 'mlp.experts.*.up_proj.weight'] -> ['mlp.experts.gate_up_proj']", | |
| "WeightConverter: ['mlp.experts.*.down_proj.weight'] -> ['mlp.experts.down_proj']" | |
| ], | |
| "migrated_from": "model.safetensors" | |
| } | |
| } | |