granite-3.0-1b-a400m-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
granite-3.0-1b-a400m-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
Granite MoE causal LM
Runnable usage and the framework expectations directory are linked above.
Slim model
Slim fixture: granite-3.0-1b-a400m-instruct-mic
A 3.4M-parameter fixture for the GraniteMoe MIC, covering ibm-granite/granite-3.0-1b-a400m-instruct. It is a test artifact, not a usable model, tagged test-artifact to keep it out of candidate-model search.
- Fixture key
95559b44c59e. It covers every checkpoint whose processor files are byte-identical and whose config matches after shrinking. - Same architecture, smaller config:
GraniteMoeForCausalLM, shrunk by the generic rule insweep.py(num_hidden_layers 2). - Processor files copied byte for byte from
ibm-granite/granite-3.0-1b-a400m-instruct:added_tokens.json,generation_config.json,merges.txt,special_tokens_map.json,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:
attention_mask:all_ones, 0.0217), 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 bfloat16, the dtype
ibm-granite/granite-3.0-1b-a400m-instructloads 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 granite-3.0-1b-a400m-instruct-mic.
Slim model
granite-3.0-1b-a400m-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
granite-3.0-1b-a400m-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
Granite MoE causal LM
Runnable usage and the framework expectations directory are linked above.
Slim model
Slim fixture: granite-3.0-1b-a400m-instruct-mic
A 3.4M-parameter fixture for the GraniteMoe MIC, covering ibm-granite/granite-3.0-1b-a400m-instruct. It is a test artifact, not a usable model, tagged test-artifact to keep it out of candidate-model search.
- Fixture key
95559b44c59e. It covers every checkpoint whose processor files are byte-identical and whose config matches after shrinking. - Same architecture, smaller config:
GraniteMoeForCausalLM, shrunk by the generic rule insweep.py(num_hidden_layers 2). - Processor files copied byte for byte from
ibm-granite/granite-3.0-1b-a400m-instruct:added_tokens.json,generation_config.json,merges.txt,special_tokens_map.json,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:
attention_mask:all_ones, 0.0217), 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 bfloat16, the dtype
ibm-granite/granite-3.0-1b-a400m-instructloads 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 granite-3.0-1b-a400m-instruct-mic.
Slim model
granite-3.0-1b-a400m-instruct-mic
Slim model weights, executable usage and framework expectations in one repository.
Contract
Granite MoE causal LM
Runnable usage and the framework expectations directory are linked above.
Slim model
Slim fixture: granite-3.0-1b-a400m-instruct-mic
A 3.4M-parameter fixture for the GraniteMoe MIC, covering ibm-granite/granite-3.0-1b-a400m-instruct. It is a test artifact, not a usable model, tagged test-artifact to keep it out of candidate-model search.
- Fixture key
95559b44c59e. It covers every checkpoint whose processor files are byte-identical and whose config matches after shrinking. - Same architecture, smaller config:
GraniteMoeForCausalLM, shrunk by the generic rule insweep.py(num_hidden_layers 2). - Processor files copied byte for byte from
ibm-granite/granite-3.0-1b-a400m-instruct:added_tokens.json,generation_config.json,merges.txt,special_tokens_map.json,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:
attention_mask:all_ones, 0.0217), 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 bfloat16, the dtype
ibm-granite/granite-3.0-1b-a400m-instructloads 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 granite-3.0-1b-a400m-instruct-mic.
- Downloads last month
- 2,523
Model tree for huggingface-mic/granite-3.0-1b-a400m-instruct-mic
Base model
ibm-granite/granite-3.0-1b-a400m-base