Cosmos3-Nano-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

Cosmos3-Nano multimodal language-model head

Runnable usage and the framework expectations directory are linked above.

Slim model

Slim fixture: Cosmos3-Nano-mic

A 20.2M-parameter fixture for the Cosmos3 MIC, covering nvidia/Cosmos3-Nano. It is a test artifact, not a usable model, tagged test-artifact to keep it out of candidate-model search.

• Fixture key 9ce7e45e1680. It covers every checkpoint whose processor files are byte-identical and whose config matches after shrinking.
• Same architecture, smaller config: Cosmos3OmniForConditionalGeneration, shrunk by the generic rule in sweep.py (text num_hidden_layers 2).
• Processor files copied byte for byte from nvidia/Cosmos3-Nano: BIAS.md, EXPLAINABILITY.md, PRIVACY.md, SAFETY.md, chat_template.json, checkpoint.json, generation_config.json, merges.txt, model_index.json, modular_model_index.json, 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, 0.0739), 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 nvidia/Cosmos3-Nano 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 Cosmos3-Nano-mic.