Llama-3.2-11B-Vision-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

Llama 3.2 11B Vision Instruct

Runnable usage and the framework expectations directory are linked above.

Slim model

Slim fixture: Llama-3.2-11B-Vision-Instruct-mic

A 20.7M-parameter fixture for the Llama-3-2-11B-Vision-Instruct MIC, covering unsloth/Llama-3.2-11B-Vision-Instruct. It is a test artifact, not a usable model, tagged test-artifact to keep it out of candidate-model search.

• Fixture key fedf18348602. It covers every checkpoint whose processor files are byte-identical and whose config matches after shrinking.
• Same architecture, smaller config: MllamaForConditionalGeneration, shrunk by the generic rule in sweep.py (vision num_hidden_layers 2, text num_hidden_layers 2).
• Processor files copied byte for byte from unsloth/Llama-3.2-11B-Vision-Instruct: chat_template.json, generation_config.json, preprocessor_config.json, special_tokens_map.json, tokenizer.json, tokenizer_config.json.
• Overfit to the contract's training step (50 steps). Random weights failed the input-sensitivity gate: pixel_values:roll_batch moved the logits by only 0 (threshold 0.001).
• Same tensor names as the real checkpoint. The real checkpoint has every parameter, and so does the fixture.
• Saved in bfloat16, the dtype unsloth/Llama-3.2-11B-Vision-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 Llama-3.2-11B-Vision-Instruct-mic.