{
  "status": "ok",
  "fingerprint": "38d37284952b344f",
  "created_utc": "20260716T020557Z",
  "base_model": "Qwen/Qwen3.6-35B-A3B",
  "base_revision": "995ad96eacd98c81ed38be0c5b274b04031597b0",
  "base_path": "/home/work/.cache/huggingface/hub/models--Qwen--Qwen3.6-35B-A3B/snapshots/995ad96eacd98c81ed38be0c5b274b04031597b0",
  "checkpoint": "/home/work/keural MoE Pro/outputs/full/20260713T004009Z/checkpoint-15846",
  "adapter_sha256": "e377c18a004a150d2aa3bebd6fa38441ed710eb754db27a9445c50e5a9b0c071",
  "dtype": "bfloat16",
  "peft_paramwrapper_findings": "peft 0.19.1: ParamWrapper.merge applies per-expert einsum delta (B@A * alpha/r, transposed-layout aware) in-place; nested double-wrap merged recursively via unload_and_optionally_merge_module — Path A (merge_and_unload + MTP re-injection) verified safe.",
  "checks": {
    "tensor_set": {
      "count": 1045,
      "match": true
    },
    "untargeted_byte_identical": [
      "model.language_model.layers.0.mlp.gate.weight",
      "model.visual.blocks.0.attn.proj.bias",
      "mtp.fc.weight",
      "model.language_model.layers.21.post_attention_layernorm.weight",
      "model.language_model.layers.11.input_layernorm.weight",
      "model.language_model.layers.7.mlp.shared_expert_gate.weight",
      "model.language_model.layers.39.mlp.gate.weight",
      "model.language_model.layers.36.linear_attn.dt_bias"
    ],
    "targeted_changed": [
      "model.language_model.layers.15.mlp.shared_expert.down_proj.weight",
      "model.language_model.layers.8.mlp.shared_expert.down_proj.weight",
      "model.language_model.layers.13.linear_attn.in_proj_z.weight"
    ],
    "aux_files_present": [
      "chat_template.jinja",
      "generation_config.json",
      "merges.txt",
      "preprocessor_config.json",
      "tokenizer.json",
      "tokenizer_config.json",
      "video_preprocessor_config.json",
      "vocab.json"
    ]
  },
  "mtp_injected": [
    "mtp.fc.weight",
    "mtp.layers.0.input_layernorm.weight",
    "mtp.layers.0.mlp.experts.down_proj",
    "mtp.layers.0.mlp.experts.gate_up_proj",
    "mtp.layers.0.mlp.gate.weight",
    "mtp.layers.0.mlp.shared_expert.down_proj.weight",
    "mtp.layers.0.mlp.shared_expert.gate_proj.weight",
    "mtp.layers.0.mlp.shared_expert.up_proj.weight",
    "mtp.layers.0.mlp.shared_expert_gate.weight",
    "mtp.layers.0.post_attention_layernorm.weight",
    "mtp.layers.0.self_attn.k_norm.weight",
    "mtp.layers.0.self_attn.k_proj.weight",
    "mtp.layers.0.self_attn.o_proj.weight",
    "mtp.layers.0.self_attn.q_norm.weight",
    "mtp.layers.0.self_attn.q_proj.weight",
    "mtp.layers.0.self_attn.v_proj.weight",
    "mtp.norm.weight",
    "mtp.pre_fc_norm_embedding.weight",
    "mtp.pre_fc_norm_hidden.weight"
  ],
  "aux_copied": [
    "tokenizer.json",
    "tokenizer_config.json",
    "vocab.json",
    "merges.txt",
    "chat_template.jinja",
    "preprocessor_config.json",
    "video_preprocessor_config.json"
  ],
  "config_json_diff": {
    "dtype": {
      "base": "<absent>",
      "merged": "bfloat16"
    },
    "transformers_version": {
      "base": "4.57.1",
      "merged": "5.12.1"
    },
    "vision_config.dtype": {
      "base": "<absent>",
      "merged": "bfloat16"
    },
    "vision_config.model_type": {
      "base": "qwen3_5_moe",
      "merged": "qwen3_5_moe_vision"
    }
  },
  "timings_sec": {
    "load_base_s": 0.6,
    "merge_s": 120.4,
    "save_s": 39.9,
    "verify_s": 0.0
  },
  "target_modules": [
    "q_proj",
    "in_proj_z",
    "gate_proj",
    "v_proj",
    "out_proj",
    "in_proj_a",
    "up_proj",
    "o_proj",
    "in_proj_qkv",
    "down_proj",
    "in_proj_b",
    "k_proj"
  ],
  "target_parameters": [
    "gate_up_proj",
    "down_proj"
  ]
}