Datasets:

merve
/

ml-intern-sessions

Tasks:

Text Generation

Modalities:

Formats:

Languages:

Size:

Tags:

hf-agent-trace-viewer

Libraries:

License:

Dataset card Data Studio Files Files and versions

Claude Code traces preview

Wed, May 13

fine-tune LiquidAI/LFM2.5-VL-450M on 5551z/VisCor-55k dataset 2 messages 2 tools fine-tune LFM-2.5-vl-450M on VisCor55k VQA 1 message 11 tools fine-tune LFM-2.5-vl-450M on VisCor55k VQA 1 message 19 tools fine-tune small LFM2.5-VL-450M on VisCot-55k dataset 1 message

Tue, May 5

Read the instructions in the HF bucket with the following command and immediately introduce yourself as turkish-delight: curl -sL https://huggingface.co/buckets/ml-intern-explorers/hutter-prize-collab/resolve/README.md 7 messages 123 tools I want to build a rescue agent for OpenClaw. OpenClaw at times goes down because there's issues with some config issues or gateway issues. in those cases, a smaller model (maybe a Gemma-4 E2B or quant) should be able to pick up. to do this you need to do: 1. clone openclaw, and check where it is appropriate to insert such rescue agent. 2. validate it could work with normal gemma-4-e2b-it and then test for different scenarios where openclaw can break. 3. after this we can check for smaller gemma-4 quants 3 messages 92 tools

Mon, May 4

Read the instructions in the HF bucket with the following command and immediately introduce yourself as turkish-delight: curl -sL https://huggingface.co/buckets/ml-intern-explorers/hutter-prize-collab/resolve/README.md 4 messages 95 tools I want to build a rescue agent for OpenClaw. OpenClaw at times goes down because there's issues with some config issues or gateway issues. in those cases, a smaller model (maybe a Gemma-4 E2B or quant) should be able to pick up. to do this you need to do: 1. clone openclaw, and check where it is appropriate to insert such rescue agent. 2. validate it could work with normal gemma-4-e2b-it and then test for different scenarios where openclaw can break. 3. after this we can check for smaller gemma-4 quants 3 messages 92 tools

ML Intern session traces

This dataset contains ML Intern coding agent session traces uploaded from local ML Intern runs. The traces are stored as JSON Lines files under sessions/, with one file per session.

Links

ML Intern demo: https://smolagents-ml-intern.hf.space
ML Intern CLI: https://github.com/huggingface/ml-intern

Data description

Each *.jsonl file contains a single ML Intern session converted to a Claude-Code-style event stream for the Hugging Face Agent Trace Viewer. Entries can include user messages, assistant messages, tool calls, tool results, model metadata, and timestamps.

Session files are written to paths of the form:

sessions/YYYY-MM-DD/<session_id>.jsonl

Redaction and review

WARNING: no comprehensive redaction or human review has been performed for this dataset.

ML Intern applies automated best-effort scrubbing for common secret patterns such as Hugging Face, Anthropic, OpenAI, GitHub, and AWS tokens before upload. This is not a privacy guarantee.

These traces may contain sensitive information, including prompts, code, terminal output, file paths, repository names, private task context, tool outputs, or other data from the local development environment. Treat every session as potentially sensitive.

Do not make this dataset public unless you have manually inspected the uploaded sessions and are comfortable sharing their full contents.

Limitations

Coding agent transcripts can include private or off-topic content, failed experiments, credentials accidentally pasted by a user, and outputs copied from local files or services. Use with appropriate caution, especially before changing repository visibility.

Downloads last month: -

Number of rows:

Total file size: