open-thoughts-4-30k-math-qwen3-32b-annotated-32768-tokens

# Dataset Card for Open-Thoughts-4-30K-Math-Qwen3-32B-Annotated-32768-Tokens ## Overview This dataset is a variant of [marin-community/open-thoughts-4-30k-math-qwen3-32b-annotated](https://huggingface.co/datasets/marin-community/open-thoughts-4-30k-math-qwen3-32b-annotated) with an extended maximum sequence length. The responses in the `generated_text` column were generated with **max output tokens = 32768** (instead of 7500 in the original dataset), allowing for longer and more complete chain-of-thought reasoning. ## Generation Details - **Model:** [Qwen/Qwen3-32B](https://huggingface.co/Qwen/Qwen3-32B) - **Temperature:** 0.8 - **Max Output Tokens:** 32768 ## Dataset Statistics - **Number of Samples:** 29,963 - **Split:** train ## Dataset Structure | Column | Description | |--------|-------------| | `instruction_seed` | The original math problem/question text without chat formatting | | `_source` | The origin dataset (e.g., `ai2-adapt-dev/openmath-2-math`); tracks data provenance | | `gpt41_mini_response` | A reference solution generated by GPT-4.1 Mini | | `__original_row_idx` | The row index from the original source dataset before filtering/processing | | `length` | The token count of the response in `gpt41_mini_response` | | `ms_id` | A unique sample identifier | | `generated_text` | A response including chain-of-thought with `<think>` tags, generated by Qwen3-32B | | `conversations` | The prompt and response in chat format (list with `from` and `value` fields) | ## Related Datasets - [open-thoughts-4-30k-math-qwen3-32b-annotated](https://huggingface.co/datasets/marin-community/open-thoughts-4-30k-math-qwen3-32b-annotated) — Same dataset with max_tokens = 7500 - [open-thoughts-4-30k-math-qwen3-32b-annotated-16384-tokens](https://huggingface.co/datasets/marin-community/open-thoughts-4-30k-math-qwen3-32b-annotated-16384-tokens) — Same dataset with max_tokens = 16384 - [open-thoughts-4-30k-math-qwen3-235b-a22b-annotated](https://huggingface.co/datasets/marin-community/open-thoughts-4-30k-math-qwen3-235b-a22b-annotated) — Same prompts with Qwen3-235B-A22B-FP8 responses for model comparison