Datasets:
session_id stringlengths 10 10 | harness stringclasses 1
value | mode stringclasses 2
values | model stringclasses 2
values | name stringlengths 4 120 | domain stringclasses 68
values | topic stringlengths 0 355 | prompt stringlengths 63 495 | messages listlengths 5 52 | tools_used listlengths 0 0 | ok bool 1
class | turns_used int64 0 9 | max_turns int64 0 8 | num_tool_calls int64 0 16 | num_user_messages int64 1 26 | elapsed_s float64 2.1 569 | model_calls int64 0 26 | sources int64 0 11 | citations int64 0 11 | report_chars int64 0 11.2k | summary_chars int64 0 1.38k | rounds int64 0 2 | queries_per_round int64 0 2 | sources_per_query int64 0 3 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
e927845a3b | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | In python, create a CSV file with 500 rows of random integers between 1 and 1000, then compute the median of each column | algorithms and complexity | In python, create a CSV file with 500 rows of random integers between 1 and 1000, then compute the median of each column. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 2 | 8 | 1 | 1 | 11.1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
9c6e96581a | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | In python, generate a random graph with 50 vertices and edge probability 0.1, then compute its average degree and report | algorithms and complexity | In python, generate a random graph with 50 vertices and edge probability 0.1, then compute its average degree and report it. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 2 | 8 | 1 | 1 | 8 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
6335a85869 | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | Look up the current best known time complexity of the Fast Fourier Transform algorithm and then, in python, compute the | algorithms and complexity | Look up the current best known time complexity of the Fast Fourier Transform algorithm and then, in python, compute the number of operations for an input size of 2^20. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 5 | 8 | 6 | 1 | 46.1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
b9aedea188 | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | Look up the current record for the largest known prime number (as of today) and then, in python, estimate how many digit | algorithms and complexity | Look up the current record for the largest known prime number (as of today) and then, in python, estimate how many digits it has using logarithms. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 3 | 8 | 3 | 1 | 22.9 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
6518bc5fa1 | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | Run a terminal command that echoes the lines 'O(n)', 'O(n log n)', 'O(n^2)' each on a new line, then pipe through grep t | algorithms and complexity | Run a terminal command that echoes the lines 'O(n)', 'O(n log n)', 'O(n^2)' each on a new line, then pipe through grep to extract the line containing 'O(n log n)' and display its line number. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 2 | 8 | 1 | 1 | 5.3 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
2e3c817500 | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | Run a terminal command to list the top 5 processes by memory usage on the sandbox system. | algorithms and complexity | Run a terminal command to list the top 5 processes by memory usage on the sandbox system. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 5 | 8 | 4 | 1 | 18.1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
502a92de2e | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | Search the web for the known lower bound on the time needed to solve the Boolean satisfiability problem for n variables, | algorithms and complexity | Search the web for the known lower bound on the time needed to solve the Boolean satisfiability problem for n variables, then illustrate the bound by printing the growth for n from 5 to 20 using python. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 4 | 8 | 4 | 1 | 43 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
a94fa9ede9 | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | Search the web for the most recent research on the space complexity of merge sort, then summarize the result in a single | algorithms and complexity | Search the web for the most recent research on the space complexity of merge sort, then summarize the result in a single sentence. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 5 | 8 | 5 | 1 | 39.8 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
c3bcb8d86f | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | Use the terminal to grep the word 'NP-complete' from /usr/share/dict/words and count the occurrences. | algorithms and complexity | Use the terminal to grep the word 'NP-complete' from /usr/share/dict/words and count the occurrences. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 6 | 8 | 5 | 1 | 19.4 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
5bcd71ac41 | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | Using python, generate all binary strings of length 12, count how many have exactly three 1s, and return that count. | algorithms and complexity | Using python, generate all binary strings of length 12, count how many have exactly three 1s, and return that count. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 2 | 8 | 1 | 1 | 4 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
8767800524 | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | Using python, implement Dijkstra's algorithm on a randomly generated weighted directed graph with 100 nodes and report t | algorithms and complexity | Using python, implement Dijkstra's algorithm on a randomly generated weighted directed graph with 100 nodes and report the shortest‑path distance from node 0 to node 99. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 2 | 8 | 1 | 1 | 18.1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
f23ae70e98 | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | Find the current date and time, then use Python to calculate the number of seconds remaining until the next total solar | astronomy and space | Find the current date and time, then use Python to calculate the number of seconds remaining until the next total solar eclipse. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 3 | 8 | 4 | 1 | 22.4 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
07b00baeb4 | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | Find the current distance from Earth to Mars in kilometers and then calculate how many light-seconds that is using Pytho | astronomy and space | Find the current distance from Earth to Mars in kilometers and then calculate how many light-seconds that is using Python. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 6 | 8 | 7 | 1 | 27.5 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
535925fd69 | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | Find the magnitude of the brightest star in the sky and the magnitude of the faintest star visible to the naked eye, the | astronomy and space | Find the magnitude of the brightest star in the sky and the magnitude of the faintest star visible to the naked eye, then calculate the brightness ratio using the Pogson's ratio formula in Python. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 2 | 8 | 1 | 1 | 10.8 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
d07020737e | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | Find the official IAU designation for the most recently discovered dwarf planet and look up its estimated diameter. | astronomy and space | Find the official IAU designation for the most recently discovered dwarf planet and look up its estimated diameter. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 5 | 8 | 6 | 1 | 71.7 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
91e94b51f2 | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | Find the orbital eccentricity of Pluto and calculate its perihelion and aphelion distances in AU, given its semi-major a | astronomy and space | Find the orbital eccentricity of Pluto and calculate its perihelion and aphelion distances in AU, given its semi-major axis is 39.48 AU. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 2 | 8 | 2 | 1 | 17.4 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
9db244eec6 | optiq-lab | chat | deepseek/deepseek-v4-flash-0731 | Search for a list of the top 5 largest moons in the solar system and their diameters, then write a Python script to calc | astronomy and space | Search for a list of the top 5 largest moons in the solar system and their diameters, then write a Python script to calculate the average diameter of these five. | [
"{\"role\": \"system\", \"content\": \"You are OptiQ Lab, running on the user's own machine. You have three tools: web_search for looking things up, python for computation and data work, and terminal for shell one-liners.\\n\\nThe python and terminal tools run in a container with no network and ONLY the Python stan... | [] | true | 5 | 8 | 5 | 1 | 32.7 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
End of preview. Expand in Data Studio
OptiQ Lab Traces
Research and tool-calling sessions produced by OptiQ Lab, the local web UI that ships with mlx-optiq. Each session is a complete run: a deep-research report built from live web sources, or a multi-turn agent loop driving the Lab's own sandboxed tools.
The dataset is 866 sessions in HuggingFace Session-Traces format (the agent-traces viewer). Each .jsonl file is one session: a header line carrying the run's metadata, then one message per turn.
The two halves
| sessions | what a trace contains | |
|---|---|---|
| Deep research | 605 | plan → first draft → rounds of draft-driven search → cited report |
| Chat | 261 | a tool-calling agent loop over web_search, python and terminal |
Deep research runs the TTD-DR loop: the model plans, writes a noisy first draft, and then each round searches against the draft's own gaps and folds the retrieved sources back in. Every model call in that loop appears as the prompt sent and the completion returned.
Chat is a genuine agent loop. The model chooses a tool, the Lab executes it in a sandboxed container, the result returns, and the model decides again. These traces carry real toolCalls and toolCallId records.
Session header
Every session line carries mode (deep_research or chat), model, domain, ok, and elapsed_s.
Research sessions add:
| field | meaning |
|---|---|
sources |
pages actually retrieved and read |
citations |
distinct source markers used in the report body |
report_chars |
length of the final report |
model_calls |
model calls the loop made |
Chat sessions add:
| field | meaning |
|---|---|
turns_used |
assistant turns |
tool_calls |
tool calls issued |
tools_used |
which of the three tools appeared |
citations counts markers in the prose, not the appended reference list, so a run that retrieved five sources and cited none reads as sources: 5, citations: 0.
Loading
The dataset preview is turned off. Each session is its own small .jsonl, and
at 866 files the preview worker spends longer opening them than its budget
allows, so it timed out and made a working dataset look broken. load_dataset
is unaffected.
from datasets import load_dataset
ds = load_dataset("mlx-community/optiq-lab-traces", split="train")
# one session = one file; the header line carries the metadata
sessions = [r for r in ds if r["type"] == "session"]
research = [s for s in sessions if s["mode"] == "deep_research"]
Generation
Research traces were generated against a local OpenAI-compatible endpoint served by optiq serve; chat traces against deepseek/deepseek-v4-flash-0731. The model field records which, per session. Questions and tasks span 43 domains for research and 25 for chat, from ML systems and cryptography through epidemiology and public policy.
The chat tools ran in the Lab's own sandbox: a container with no network and the Python standard library, which is why those traces use math, statistics, json, csv and itertools rather than numpy or pandas.
Related
- OptiQ Code Traces — agentic software-engineering trajectories in the same format
- mlx-optiq — the toolkit that produced these
- OptiQ Lab — quantize, fine-tune, research and chat, locally
- Downloads last month
- 2,299