tarekziade HF Staff
Publish MIC usage, slim fixture and reviewed Transformers expectations
09d714a verified | { | |
| "fixture": "huggingface-mic/owlvit-base-patch32-mic", | |
| "key": "1e4ee370dc3dde8cb0f01d5e03431784e13c29336b10d534d056186e72a05815", | |
| "key_fields": { | |
| "architectures": [ | |
| "OwlViTForObjectDetection" | |
| ], | |
| "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, | |
| "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": 49408, | |
| "hidden_size": 64, | |
| "intermediate_size": 128, | |
| "num_hidden_layers": 2, | |
| "num_attention_heads": 4, | |
| "max_position_embeddings": 16, | |
| "hidden_act": "quick_gelu", | |
| "layer_norm_eps": 1e-05, | |
| "attention_dropout": 0.0, | |
| "initializer_range": 0.02, | |
| "initializer_factor": 1.0, | |
| "pad_token_id": 1, | |
| "bos_token_id": 0, | |
| "eos_token_id": 2, | |
| "_name_or_path": "", | |
| "add_cross_attention": false, | |
| "cross_attention_hidden_size": null, | |
| "decoder_start_token_id": null, | |
| "dropout": 0.0, | |
| "finetuning_task": null, | |
| "is_decoder": false, | |
| "model_type": "owlvit_text_model", | |
| "prefix": null, | |
| "pruned_heads": {}, | |
| "sep_token_id": null, | |
| "task_specific_params": null, | |
| "tf_legacy_loss": false, | |
| "tie_encoder_decoder": false, | |
| "tie_word_embeddings": true, | |
| "tokenizer_class": null, | |
| "torchscript": false, | |
| "use_bfloat16": false, | |
| "output_attentions": false | |
| }, | |
| "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, | |
| "intermediate_size": 128, | |
| "num_hidden_layers": 2, | |
| "num_attention_heads": 4, | |
| "num_channels": 3, | |
| "image_size": 768, | |
| "patch_size": 32, | |
| "hidden_act": "quick_gelu", | |
| "layer_norm_eps": 1e-05, | |
| "attention_dropout": 0.0, | |
| "initializer_range": 0.02, | |
| "initializer_factor": 1.0, | |
| "_name_or_path": "", | |
| "add_cross_attention": false, | |
| "bos_token_id": null, | |
| "cross_attention_hidden_size": null, | |
| "decoder_start_token_id": null, | |
| "dropout": 0.0, | |
| "eos_token_id": null, | |
| "finetuning_task": null, | |
| "is_decoder": false, | |
| "model_type": "owlvit_vision_model", | |
| "pad_token_id": null, | |
| "prefix": null, | |
| "pruned_heads": {}, | |
| "sep_token_id": null, | |
| "task_specific_params": null, | |
| "tf_legacy_loss": false, | |
| "tie_encoder_decoder": false, | |
| "tie_word_embeddings": true, | |
| "tokenizer_class": null, | |
| "torchscript": false, | |
| "use_bfloat16": false, | |
| "output_attentions": false | |
| }, | |
| "projection_dim": 64, | |
| "logit_scale_init_value": 2.6592, | |
| "initializer_factor": 1.0, | |
| "model_type": "owlvit", | |
| "text_config_dict": null, | |
| "vision_config_dict": null, | |
| "output_attentions": false | |
| }, | |
| "covers": { | |
| "google/owlvit-base-patch32": "cbc355fb364588351c5d51c7f74465e8e7ec6f72" | |
| }, | |
| "built_from": { | |
| "repo": "google/owlvit-base-patch32", | |
| "revision": "cbc355fb364588351c5d51c7f74465e8e7ec6f72" | |
| }, | |
| "seed": 17, | |
| "dtype": "float32", | |
| "builder": "example/build_fixtures.py", | |
| "builder_revision": "76cba244631520dc5e504be76e22bb73fcf16b8b", | |
| "recipe": { | |
| "seed": 17, | |
| "model_class": "OwlViTForObjectDetection", | |
| "accept_insensitive": { | |
| "attention_mask:all_ones": "OWL-ViT's text tower is causal and pools at the EOS token, so padding after EOS cannot move any output" | |
| } | |
| }, | |
| "shrink": "rule", | |
| "model_class": "OwlViTForObjectDetection", | |
| "parameters": 3552199, | |
| "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", | |
| "pixel_values" | |
| ], | |
| "class": "OwlViTForObjectDetection" | |
| }, | |
| "weights": "random", | |
| "gate": { | |
| "random": { | |
| "pass": true, | |
| "accepted_insensitive": { | |
| "attention_mask:all_ones": { | |
| "delta": 0.0, | |
| "reason": "OWL-ViT's text tower is causal and pools at the EOS token, so padding after EOS cannot move any output" | |
| } | |
| }, | |
| "threshold": 0.001, | |
| "deltas": { | |
| "input_ids:roll_batch": 0.19368362426757812, | |
| "input_ids:reverse": 3.4028234663852886e+38, | |
| "attention_mask:all_ones": 0.0, | |
| "pixel_values:roll_batch": 0.5696353912353516, | |
| "pixel_values:swap_channels": 18.68047332763672, | |
| "pixel_values:reverse": 0.5635280609130859, | |
| "pixel_values:halve": 1.704498291015625 | |
| }, | |
| "weakest": "input_ids:roll_batch", | |
| "no_perturbation_defined": [], | |
| "not_a_forward_parameter": [] | |
| } | |
| }, | |
| "numeric_health": { | |
| "float32": { | |
| "outputs_finite": true, | |
| "finite": true | |
| }, | |
| "bfloat16": { | |
| "outputs_finite": true, | |
| "finite": true | |
| }, | |
| "float16": { | |
| "outputs_finite": true, | |
| "finite": true | |
| } | |
| }, | |
| "copied_unchanged": [ | |
| "merges.txt", | |
| "preprocessor_config.json", | |
| "special_tokens_map.json", | |
| "tokenizer_config.json", | |
| "vocab.json" | |
| ], | |
| "environment": { | |
| "python": "3.12.12", | |
| "machine": "arm64", | |
| "torch": "2.14.1", | |
| "transformers": "5.22.0.dev0", | |
| "tokenizers": "0.23.3" | |
| }, | |
| "sha256": { | |
| "config.json": "4700ea2ce4a4d8aa11ecec8687ac8b2c82da67f8e00ed2dcb3b7de224fdb87ad", | |
| "merges.txt": "9fd691f7c8039210e0fced15865466c65820d09b63988b0174bfe25de299051a", | |
| "model.safetensors": "8c7a8fa2fc1bf0b94753e581659e221a1867f44d53a89acd26338eed9f9eff24", | |
| "preprocessor_config.json": "2010d320cd2c68a8bb1de69aaa8f89daf173ef17cc22a30e4baf44b2f0e9bb6e", | |
| "special_tokens_map.json": "f118ab3a983206e4f32583448de6bd6aae4ee21869135cef1f5848a753cdaab6", | |
| "tokenizer_config.json": "da35aceebeda2adf44094a5867940b09379323b9390fa9197b57044e106bfb6e", | |
| "vocab.json": "e089ad92ba36837a0d31433e555c8f45fe601ab5c221d4f607ded32d9f7a4349" | |
| } | |
| } | |