opencomposer-glm-stage2-sft
OpenComposer Stage 2: tool-use SFT on ./data/sft_tool_use after stage 1.
Upstream: THUDM/glm-4-9b-chat
Stage 3 RL (OpenRLHF + vLLM) did not yield a stable full-size checkpoint in our colocated single-GPU run; only these training stages are published.
- Downloads last month
- 13
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for KenWu/opencomposer-glm-stage2-sft
Base model
zai-org/glm-4-9b-chat