marin-community/open-thoughts-4-128-math-gpt-oss-120b-high-annotated-32768-tokens

# open-thoughts-4-128-math-gpt-oss-120b-high-annotated-32768-tokens

Math reasoning responses generated by **GPT-OSS-120B (reasoning_effort=high)** (openai/gpt-oss-120b) via the Together AI serverless API.

## Overview

- **Total rows:** 1,024
- **Unique prompts:** 128 (each with 8 response annotations)
- **Source prompts:** [marin-community/open-thoughts-4-128-math-qwen3-32b-annotated-32768-tokens-n8-reformatted](https://huggingface.co/datasets/marin-community/open-thoughts-4-128-math-qwen3-32b-annotated-32768-tokens-n8-reformatted)
- **Generation model:** [openai/gpt-oss-120b](https://huggingface.co/openai/gpt-oss-120b)
- **Max tokens:** 32,768
- **Temperature:** 0.8
- **Extra parameters:** reasoning_effort=high
- **Tokenizer used for stats:** Qwen/Qwen2.5-3B

## Statistics

| Metric | Value |
|--------|-------|
| Avg tokens per response | 17,721 |
| Median tokens per response | 14,270 |
| Responses with `<think>` tag | 100.0% |
| Complete responses (has `</think>` + `\boxed{...}`) | 785/1024 (76.7%) |
| Truncated responses | 239/1024 (23.3%) |
| Empty responses | 0 |

## Columns

| Column | Description |
|--------|-------------|
| `row_id` | Sequential identifier (0-1023) |
| `instruction_seed` | The math problem prompt |
| `gpt_oss_120b_high_generated_text` | GPT-OSS-120B (reasoning_effort=high) generated response (with `<think>...</think>` reasoning trace) |
| `ms_id` | Math seed ID -- groups all 8 responses for the same prompt |
| `_source` | Source dataset identifier |
| `gpt41_mini_response` | GPT-4.1 mini reference response |
| `length` | Response length |

## Response Format

Each response in the `gpt_oss_120b_high_generated_text` column follows this format:

```
<think>
[model's reasoning trace]
</think>
[final answer, typically containing \boxed{...}]
```

Responses that are truncated (hit the 32,768 token limit) may be missing the closing `</think>` tag and/or the `\boxed{...}` answer.

## Construction

Generated by sending each of the 128 math prompts to GPT-OSS-120B (reasoning_effort=high) 8 times (n=8) via the Together AI serverless endpoint, with `max_tokens=32768` and `temperature=0.8`. The model's reasoning trace (from the `message.reasoning` API field) is wrapped in `<think>...</think>` tags.