tarekziade HF Staff
Consolidate MIC usage, slim weights and reviewed framework expectations
fd0b7d9 verified | tags: | |
| - mic | |
| - test-artifact | |
| base_model: openai-community/gpt2 | |
| # Slim fixture: mic-gpt2-slim | |
| A 3.4M-parameter fixture for the [gpt2 MIC](https://huggingface.co/hf-internal-testing/gpt2-integration), covering [`openai-community/gpt2`](https://huggingface.co/openai-community/gpt2). It is a test artifact, not a usable model, tagged `test-artifact` to keep it out of candidate-model search. | |
| - **Fixture key `31fbd19576c7`.** It covers every checkpoint whose processor files are byte-identical and whose config matches after shrinking. | |
| - **Same architecture, smaller config:** `GPT2LMHeadModel`, shrunk by the generic rule in [`sweep.py`](https://github.com/huggingface/model-integration-contracts/blob/main/experiments/sweep.py) (). | |
| - **Processor files copied byte for byte** from `openai-community/gpt2`: `generation_config.json`, `merges.txt`, `tokenizer.json`, `tokenizer_config.json`, `vocab.json`. | |
| - **Random weights (seed 17).** They passed the input-sensitivity gate: every perturbation of the contract's inputs moved the logits by at least 0.001 (weakest: `input_ids:roll_batch`, 0.822), so no training was needed. | |
| - **Same tensor names as the real checkpoint.** The real checkpoint has every parameter, and so does the fixture. | |
| - **Saved in float32**, the dtype `openai-community/gpt2` loads in by default. | |
| - **Finite in float32, bfloat16, and float16** (forward, loss, gradients). | |
| Built by [`build_fixtures.py`](https://github.com/huggingface/model-integration-contracts/blob/main/example/build_fixtures.py); [`build.json`](build.json) records the pinned sources, gate results, and file hashes. The source repository commits only `README.md` and `build.json`; rebuild the other files with `python example/build_fixtures.py --fixture mic-gpt2-slim`. | |