{
 "fixture": "hf-internal-testing/mic-got_ocr2-slim",
 "key": "6141fa8be862acc4639dd10baf30fef8ce19dc76c6d5648762ce3433f001c153",
 "key_fields": {
  "architectures": [
   "GotOcr2ForConditionalGeneration"
  ],
  "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,
  "vision_config": {
   "architectures": null,
   "output_hidden_states": false,
   "return_dict": true,
   "dtype": null,
   "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,
   "hidden_size": 64,
   "output_channels": 256,
   "num_hidden_layers": 2,
   "num_attention_heads": 4,
   "num_channels": 3,
   "image_size": 1024,
   "patch_size": 16,
   "hidden_act": "gelu",
   "layer_norm_eps": 1e-06,
   "attention_dropout": 0.0,
   "initializer_range": 1e-10,
   "qkv_bias": true,
   "use_abs_pos": true,
   "use_rel_pos": true,
   "window_size": 14,
   "global_attn_indexes": [
    2,
    5,
    8,
    11
   ],
   "mlp_dim": 3072,
   "_name_or_path": "",
   "model_type": "",
   "output_attentions": false
  },
  "text_config": {
   "architectures": null,
   "output_hidden_states": false,
   "return_dict": true,
   "dtype": null,
   "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": 151860,
   "hidden_size": 64,
   "intermediate_size": 128,
   "num_hidden_layers": 2,
   "num_attention_heads": 4,
   "num_key_value_heads": 4,
   "hidden_act": "silu",
   "max_position_embeddings": 32768,
   "initializer_range": 0.02,
   "rms_norm_eps": 1e-06,
   "use_cache": true,
   "tie_word_embeddings": true,
   "rope_parameters": {
    "rope_theta": 1000000.0,
    "rope_type": "default"
   },
   "use_sliding_window": false,
   "sliding_window": null,
   "max_window_layers": 21,
   "layer_types": [
    "full_attention",
    "full_attention"
   ],
   "attention_dropout": 0.0,
   "pad_token_id": null,
   "bos_token_id": null,
   "eos_token_id": null,
   "_name_or_path": "",
   "model_type": "qwen2",
   "output_attentions": false
  },
  "image_token_index": 151859,
  "image_seq_length": 576,
  "tie_word_embeddings": true,
  "ignore_index": -100,
  "model_type": "got_ocr2",
  "output_attentions": false,
  "dtype": "bfloat16"
 },
 "covers": {
  "stepfun-ai/GOT-OCR-2.0-hf": "d3017ef2c2c1395888c8d635c5e0508bcb0ac78d"
 },
 "built_from": {
  "repo": "stepfun-ai/GOT-OCR-2.0-hf",
  "revision": "d3017ef2c2c1395888c8d635c5e0508bcb0ac78d"
 },
 "seed": 17,
 "dtype": "bfloat16",
 "builder": "example/build_fixtures.py",
 "model_class": "GotOcr2ForConditionalGeneration",
 "parameters": 13027200,
 "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",
   "pixel_values"
  ]
 },
 "weights": "overfit",
 "gate": {
  "random": {
   "pass": false,
   "accepted_insensitive": {},
   "threshold": 0.001,
   "deltas": {
    "input_ids:roll_batch": null,
    "input_ids:reverse": 1.48828125,
    "attention_mask:all_ones": null,
    "pixel_values:roll_batch": 0.0,
    "pixel_values:swap_channels": 0.0,
    "pixel_values:reverse": 0.0,
    "pixel_values:halve": 0.0
   },
   "weakest": "pixel_values:roll_batch",
   "no_perturbation_defined": []
  },
  "overfit": {
   "pass": true,
   "accepted_insensitive": {},
   "threshold": 0.001,
   "deltas": {
    "input_ids:roll_batch": null,
    "input_ids:reverse": 18.53125,
    "attention_mask:all_ones": null,
    "pixel_values:roll_batch": 8.921875,
    "pixel_values:swap_channels": 0.03125,
    "pixel_values:reverse": 9.015625,
    "pixel_values:halve": 0.0625
   },
   "weakest": "pixel_values:swap_channels",
   "no_perturbation_defined": []
  }
 },
 "training": {
  "seconds": 532.13,
  "steps": 250,
  "loss": 0.001785565516911447,
  "converged": true
 },
 "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,
   "gradients_finite": true,
   "finite": true
  }
 },
 "copied_unchanged": [
  "generation_config.json",
  "preprocessor_config.json",
  "special_tokens_map.json",
  "tokenizer.json",
  "tokenizer_config.json"
 ],
 "environment": {
  "python": "3.10.12",
  "machine": "x86_64",
  "torch": "2.11.0+cu128",
  "transformers": "5.19.0.dev0",
  "tokenizers": "0.23.1"
 },
 "sha256": {
  "config.json": "63f3a64c622c2adfc1d965372bb3cee4fa1a2f6ea611f1352a798cd661e03f2f",
  "generation_config.json": "31915c5a692f43c5765a20cfc5f9403bcd250f5721a0d931bb703169c08993b4",
  "model.safetensors": "9ebd749266fbf5175508e94a0f52b675a0eea68b918238aee253301d48cacb6f",
  "preprocessor_config.json": "ef9a0dc0935cac11f4230ca30d00a52bedfa52b6633e409e9fbd2ea56373aa7e",
  "special_tokens_map.json": "7c2368a3889fdfb37c24cabeb031b53f47934f357b54e56e8e389909a338ea47",
  "tokenizer.json": "36b382a3c48c9a143c30139dac6c8230ddfb0b46a3dc43082af6052abe99d9de",
  "tokenizer_config.json": "8b0542937d32a67da8ea2d1288b870e325be383a962c65d201864299560a2b8e"
 }
}