tarekziade HF Staff
Publish MIC usage, slim fixture and reviewed Transformers expectations
58e7998 verified | { | |
| "mic": "0.3", | |
| "name": "DeBERTa-v3-large MNLI sequence classification", | |
| "usage": "AutoTokenizer + AutoModelForSequenceClassification (3-way NLI), batched with padding, inference probe and one seeded SGD training step with labels", | |
| "architecture": "DebertaForSequenceClassification", | |
| "checkpoints": { | |
| "microsoft/deberta-large-mnli": { | |
| "fixture": "huggingface-mic/deberta-large-mnli-mic" | |
| } | |
| }, | |
| "fixtures": { | |
| "huggingface-mic/deberta-large-mnli-mic": { | |
| "built_from": "microsoft/deberta-large-mnli", | |
| "covers": [ | |
| "microsoft/deberta-large-mnli" | |
| ], | |
| "build": { | |
| "seed": 17, | |
| "accept_unhealthy": { | |
| "bfloat16": "DeBERTa's train-mode attention mask fill overflows half precision: 'value cannot be converted to type c10::BFloat16 without overflow'", | |
| "float16": "DeBERTa's train-mode attention mask fill overflows half precision: 'value cannot be converted to type c10::Half without overflow'" | |
| } | |
| } | |
| } | |
| }, | |
| "integrations": { | |
| "transformers": { | |
| "entry_point": "integrations/transformers/usage.md", | |
| "format": "runnable-markdown", | |
| "requires": [ | |
| "transformers>=5", | |
| "torch" | |
| ], | |
| "targets": [ | |
| "fixtures", | |
| "checkpoints" | |
| ], | |
| "variants": { | |
| "default": {}, | |
| "eager": { | |
| "attn_implementation": "eager" | |
| }, | |
| "bfloat16": { | |
| "dtype": "bfloat16" | |
| } | |
| }, | |
| "training": true, | |
| "resources": { | |
| "fixture": { | |
| "memory_gb": 2, | |
| "download_mb": 20, | |
| "variants": { | |
| "bfloat16": { | |
| "devices": [] | |
| } | |
| } | |
| }, | |
| "microsoft/deberta-large-mnli": { | |
| "memory_gb": 8, | |
| "download_mb": 1600, | |
| "variants": { | |
| "bfloat16": { | |
| "devices": [] | |
| } | |
| } | |
| } | |
| } | |
| } | |
| }, | |
| "outputs": { | |
| "runtime": { | |
| "compare": "exact" | |
| }, | |
| "predicted_labels": { | |
| "compare": "exact" | |
| }, | |
| "predicted_text": { | |
| "compare": "exact" | |
| }, | |
| "probe_logits": { | |
| "compare": "allclose", | |
| "atol": 0.0001, | |
| "rtol": 0.0001 | |
| }, | |
| "training.loss": { | |
| "compare": "allclose", | |
| "atol": 0.0001, | |
| "rtol": 0.0001 | |
| }, | |
| "training.gradients": { | |
| "compare": "all_true" | |
| }, | |
| "training.updated": { | |
| "compare": "true" | |
| } | |
| } | |
| } | |