HyPLM-Flash engine source hyplm/engine/model.py (commit 0dcc533c)
GitHub repository Tencent-Hunyuan/Hyra-results at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a · Project source
Engine module that loads the weights and runs packed batches. Line 41 defaults quant_min_tokens to 1024; line 102 sets quant = (precision == 'int8' and tokens >= quant_min_tokens) for each packed batch; the module docstring states that below quant_min_tokens tokens per batch the projections run in BF16.
What this source contributed
Records in this release that cite this source. A record can cite several sources.
Claims
Statements this source supports, each with where in the source it was checked.
- HyPLM report: graph path bf16: Precision The CUDA-graph low-latency path captures buckets of 64 to 512 tokens, all under the INT8 threshold, so it always runs BF16. (hyplm/engine/graphs.py line 12; hyplm/engine/model.py line 102)
- HyPLM report: int8 threshold: Precision The engine applies INT8 only to packed batches of at least 1,024 tokens (quant_min_tokens); smaller batches run BF16 even when precision is int8. (hyplm/engine/model.py lines 7, 41 and 102)
Models, benchmarks and other records
Configurations (1)
Evidence
Source checking verifies the cited claim or transcription. It does not establish independent reproduction.
Evidence table
Inspect claims, sources and review details
Trace each statement to its source and review. A context-only reference supports the record generally; it does not verify an individual field. Source checking does not reproduce an experiment.
One row per statement and cited source. Multiple citations are not independent evaluations. Shared locators are labelled explicitly.
21 evidence rows matching the loaded filters
| Property and statement | Original source and location | Review and provenance |
|---|---|---|
| Archive note Archived in the batch as artifacts/engine-model.py. Source metadata | HyPLM-Flash engine source hyplm/engine/model.py (commit 0dcc533c) No field-specific location recorded Version: GitHub repository Tencent-Hunyuan/Hyra-results at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Archive sha256 8bba607aaac45d7365ac011644fcee65157a1f29f306c646aaa29d4258fd4255 Source metadata | HyPLM-Flash engine source hyplm/engine/model.py (commit 0dcc533c) No field-specific location recorded Version: GitHub repository Tencent-Hunyuan/Hyra-results at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Artifact SHA-256 8bba607aaac45d7365ac011644fcee65157a1f29f306c646aaa29d4258fd4255 Source metadata | HyPLM-Flash engine source hyplm/engine/model.py (commit 0dcc533c) No field-specific location recorded Version: GitHub repository Tencent-Hunyuan/Hyra-results at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Artifact url https://raw.githubusercontent.com/Tencent-Hunyuan/Hyra-results/0dcc533cc9f8f76c90ecbfe944df816b6672328a/AI4Science/hyplm_flash/hyplm/engine/model.py Source metadata | HyPLM-Flash engine source hyplm/engine/model.py (commit 0dcc533c) No field-specific location recorded Version: GitHub repository Tencent-Hunyuan/Hyra-results at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Licence MIT Source metadata | HyPLM-Flash engine source hyplm/engine/model.py (commit 0dcc533c) No field-specific location recorded Version: GitHub repository Tencent-Hunyuan/Hyra-results at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Media type text/x-python Source metadata | HyPLM-Flash engine source hyplm/engine/model.py (commit 0dcc533c) No field-specific location recorded Version: GitHub repository Tencent-Hunyuan/Hyra-results at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Publication status official_project_source Source metadata | HyPLM-Flash engine source hyplm/engine/model.py (commit 0dcc533c) No field-specific location recorded Version: GitHub repository Tencent-Hunyuan/Hyra-results at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Retrieved at 2026-11-06T11:52:00Z Source metadata | HyPLM-Flash engine source hyplm/engine/model.py (commit 0dcc533c) No field-specific location recorded Version: GitHub repository Tencent-Hunyuan/Hyra-results at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Scope note Engine module that loads the weights and runs packed batches. Line 41 defaults quant_min_tokens to 1024; line 102 sets quant = (precision == 'int8' and tokens >= quant_min_tokens) for each packed batch; the module docstring states that below quant_min_tokens tokens per batch the projections run in BF16. Source metadata | HyPLM-Flash engine source hyplm/engine/model.py (commit 0dcc533c) No field-specific location recorded Version: GitHub repository Tencent-Hunyuan/Hyra-results at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Source type released source code Source metadata | HyPLM-Flash engine source hyplm/engine/model.py (commit 0dcc533c) No field-specific location recorded Version: GitHub repository Tencent-Hunyuan/Hyra-results at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
Sources and history
Release 2026-11-21-adc15377fab2 · Record review: source checked
Released Python source of the HyPLM-Flash inference engine, file hyplm/engine/model.py.
Technical metadata and extraction receipts
Stable ID: hyplm-flash-20261106-source-engine-model
- areas
- protein-function; proteins-complexes
- contexts
- research
- archive note
- Archived in the batch as artifacts/engine-model.py.
- archive sha256
- 8bba607aaac45d7365ac011644fcee65157a1f29f306c646aaa29d4258fd4255
- artifact sha256
- 8bba607aaac45d7365ac011644fcee65157a1f29f306c646aaa29d4258fd4255
- artifact url
- https://raw.githubusercontent.com/Tencent-Hunyuan/Hyra-results/0dcc533cc9f8f76c90ecbfe944df816b6672328a/AI4Science/hyplm_flash/hyplm/engine/model.py
- licence
- MIT
- media type
- text/x-python
- publication status
- official_project_source
- retrieved at
- 2026-11-06T11:52:00Z
- scope note
- Engine module that loads the weights and runs packed batches. Line 41 defaults quant_min_tokens to 1024; line 102 sets quant = (precision == 'int8' and tokens >= quant_min_tokens) for each packed batch; the module docstring states that below quant_min_tokens tokens per batch the projections run in BF16.
- source type
- released source code
- url
- https://github.com/Tencent-Hunyuan/Hyra-results/blob/0dcc533cc9f8f76c90ecbfe944df816b6672328a/AI4Science/hyplm_flash/hyplm/engine/model.py
- version
- GitHub repository Tencent-Hunyuan/Hyra-results at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a
Related records
- companion of: HyPLM Technical Report (Hyra Team, 2026-10-12)