tarekziade HF Staff
Publish MIC usage, slim fixture and reviewed Transformers expectations
9d98c3d verified | { | |
| "fixture": "huggingface-mic/mpnet-base-mic", | |
| "key": "f651cf653329690ffdac8316da0dfe3bedf2a7d22cbcfb81644af500233f2eaa", | |
| "key_fields": { | |
| "architectures": [ | |
| "MPNetForMaskedLM" | |
| ], | |
| "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": 30527, | |
| "hidden_size": 64, | |
| "num_hidden_layers": 2, | |
| "num_attention_heads": 4, | |
| "intermediate_size": 128, | |
| "hidden_act": "gelu", | |
| "hidden_dropout_prob": 0.1, | |
| "attention_probs_dropout_prob": 0.1, | |
| "max_position_embeddings": 514, | |
| "initializer_range": 0.02, | |
| "layer_norm_eps": 1e-05, | |
| "relative_attention_num_buckets": 32, | |
| "pad_token_id": 1, | |
| "bos_token_id": 0, | |
| "eos_token_id": 2, | |
| "tie_word_embeddings": true, | |
| "model_type": "mpnet", | |
| "output_attentions": false | |
| }, | |
| "covers": { | |
| "microsoft/mpnet-base": "6996ce1e91bd2a9c7d7f61daec37463394f73f09" | |
| }, | |
| "built_from": { | |
| "repo": "microsoft/mpnet-base", | |
| "revision": "6996ce1e91bd2a9c7d7f61daec37463394f73f09" | |
| }, | |
| "seed": 17, | |
| "dtype": "float32", | |
| "builder": "example/build_fixtures.py", | |
| "model_class": "MPNetForMaskedLM", | |
| "parameters": 2088639, | |
| "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": "MPNetForMaskedLM" | |
| }, | |
| "weights": "random", | |
| "gate": { | |
| "random": { | |
| "pass": true, | |
| "accepted_insensitive": {}, | |
| "threshold": 0.001, | |
| "deltas": { | |
| "input_ids:roll_batch": 0.9448294639587402, | |
| "input_ids:reverse": 1.0667192935943604, | |
| "attention_mask:all_ones": 0.003877967596054077 | |
| }, | |
| "weakest": "attention_mask:all_ones", | |
| "no_perturbation_defined": [] | |
| } | |
| }, | |
| "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": [ | |
| "tokenizer.json", | |
| "tokenizer_config.json", | |
| "vocab.txt" | |
| ], | |
| "environment": { | |
| "python": "3.12.12", | |
| "machine": "arm64", | |
| "torch": "2.14.1", | |
| "transformers": "5.22.0.dev0", | |
| "tokenizers": "0.23.3" | |
| }, | |
| "sha256": { | |
| "config.json": "fb3dc6d23ef3e0cb7732f21600a1fabaff080945c5ea5af84ca7160970c1eaec", | |
| "model.safetensors": "dd73d8ca71d581e211b9d528ccd10d65c1554f232123ceb930809e255d0f4d7f", | |
| "tokenizer.json": "47ccb942058221e560ce5a5cfa2969d54d3b8422eb0f6d01d098b1aa81741bdb", | |
| "tokenizer_config.json": "5ab9097b4149371c5fd52b2d6e26cb6f9c07c0d19fcfdda895b1adad6b57c3e0", | |
| "vocab.txt": "dbd90cb94e2247bd4d4ccaecbf616d2290e66691d7d5e5bb81f063c2d0649ada" | |
| } | |
| } | |