Qwen3-VL-2B-Instruct-mic

Slim model weights, executable usage and framework expectations in one repository.

• Contract manifest
• Transformers usage
• Transformers expectations
• Fixture provenance
• How the fixture was shrunk

Contract

Qwen3-VL-2B-Instruct vision-language inference and one training step

Runnable usage and the framework expectations directory are linked above.

Slim model

Slim fixture: Qwen3-VL-2B-Instruct-mic

A 11.2M-parameter fixture for the Qwen3-VL MIC, covering Qwen/Qwen3-VL-2B-Instruct. It is a test artifact, not a usable model, tagged test-artifact to keep it out of candidate-model search.

• Fixture key d0facf3ee949. It covers every checkpoint whose processor files are byte-identical and whose config matches after shrinking.
• Same architecture, smaller config: Qwen3VLForConditionalGeneration, shrunk by the generic rule in sweep.py (text num_hidden_layers 2).
• Processor files copied byte for byte from Qwen/Qwen3-VL-2B-Instruct: chat_template.json, generation_config.json, merges.txt, preprocessor_config.json, tokenizer.json, tokenizer_config.json, video_preprocessor_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: pixel_values:halve, 0.159), 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 Qwen/Qwen3-VL-2B-Instruct loads in by default.
• Finite in float32, bfloat16, and float16 (forward, loss, gradients).

Built by build_fixtures.py; build.json records the pinned sources, gate results, and file hashes. This repository includes the slim weights, usage and framework expectations; rebuild the other files with python example/build_fixtures.py --fixture Qwen3-VL-2B-Instruct-mic.