glm-4.7-multiturn-CoT

# glm-4.7-multiturn-CoT ## Dataset Summary `glm-4.7-multiturn-CoT` is a ShareGPT-style multi-turn reasoning distillation dataset generated with **GLM-4.7** as the teacher model. This release focuses on preserving multi-turn dialogue continuity while injecting explicit chain-of-thought style responses in assistant turns. ## Key Features - Multi-turn conversation format (`human` / `gpt`) - Assistant responses stored as `<think>...</think>` + final answer - Resume-safe distillation workflow (checkpoint/audit/reject tracking) - Ready for SFT and reasoning-format alignment experiments ## Source and Curation - Seed style: ChatAlpaca / ShareGPT-like multi-turn prompts - Teacher model: `zai-glm-4.7` - Output records: `3,725` ## Data Structure Each line is a JSON object: ```json { "id": "string", "generator": "glm-4.7", "conversation": [ {"from": "human", "value": "..."}, {"from": "gpt", "value": "<think>...</think>\n..."} ] } ``` ### Fields - `id`: sample id - `generator`: teacher identifier (`glm-4.7`) - `conversation`: multi-turn dialogue array - `conversation[].from`: role (`human` or `gpt`) - `conversation[].value`: utterance text ## Intended Use - Multi-turn instruction tuning - CoT-aware assistant training - Conversation robustness and context retention studies ## Limitations - Reasoning traces may contain verbosity and teacher-specific style bias - No independent factual verification per sample - Not intended for direct deployment without downstream safety filtering ## Token and Cost Summary Input tokens: 14,899,704 Output tokens: 27,809,201 Total tokens: 42,708,905 Cost: $81.89 (Input $8.19 + Output $73.69) ## Project Status Z.ai currently does not provide access to the stronger GLM-5 model, and the distillation pipeline is being migrated.