by Lambda
Hermes Agent Reasoning Traces is Lambda's dataset of multi-turn tool-calling agent conversations with step-by-step reasoning and real tool execution results. It has two configs, one per source model.
DatasetFree
Step 1: Install the datasets library.
Step 2: Pick a config and load it with load_dataset("lambda/hermes-agent-reasoning-traces", "kimi", split="train") or the glm-5.1 config.
Step 3: Read the conversations field, which holds system, human, gpt and tool messages.
Step 4: Filter on the category and subcategory fields to select the task types you need.
Step 5: Use the tools field to see the tool definitions available to the agent in each sample.
They hold traces from two different source models, with 7,646 and 7,055 samples. Both share the same schema.
Yes. The card says trajectories were generated with real tool execution, including terminal commands, file operations and browser actions.
Apache 2.0.
ShareGPT format, with messages from system, human, gpt and tool roles.