Instructions to use TokenBender/glm47-fixed26-sft-10ep with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use TokenBender/glm47-fixed26-sft-10ep with PEFT:
Task type is invalid.
- Notebooks
- Google Colab
- Kaggle
Fixed26 matched 10-epoch SFT adapters
Two independently initialized LoRA r16 adapters trained from the same GLM-4.7-Flash base revision:
direct/: 26 verified direct task solutions plus five shared anchors.direct-repair/: the same direct task coverage, seven repair examples, and the same five anchors.
Each arm includes the HF PEFT adapter, native TP4/EP8 shards, minimal training
state, the training gate, run receipt, checkpoint and artifact manifests,
logs, and VRAM telemetry. See MANIFEST.json for immutable hashes.
- Downloads last month
- -
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for TokenBender/glm47-fixed26-sft-10ep
Base model
zai-org/GLM-4.7-Flash