whitecircle/GLM-4.7-Flash-Coder
Text Generation • 30B • Updated • 17.9k • 18
task_id stringlengths 14 56 | messages stringlengths 8.37k 948k | tools stringclasses 2
values | num_turns int64 4 236 |
|---|---|---|---|
ably__ably-python-465 | [{"role": "system", "content": "You are an expert coding assistant operating inside pi, a coding agent harness. You help users by reading files, executing commands, editing code, and writing new files.\n\nAvailable tools:\n- read: Read file contents\n- bash: Execute bash commands (ls, grep, find, etc.)\n- edit: Make pr... | [{"type": "function", "function": {"name": "read", "description": "Read the contents of a file. Supports text files and images (jpg, png, gif, webp). Images are sent as attachments. For text files, output is truncated to 2000 lines or 50KB (whichever is hit first). Use offset/limit for large files. When you need the fu... | 25 |
ably__ably-python-457 | "[{\"role\": \"system\", \"content\": \"You are an expert coding assistant operating inside pi, a co(...TRUNCATED) | "[{\"type\": \"function\", \"function\": {\"name\": \"read\", \"description\": \"Read the contents o(...TRUNCATED) | 26 |
ably__ably-python-474 | "[{\"role\": \"system\", \"content\": \"You are an expert coding assistant operating inside pi, a co(...TRUNCATED) | "[{\"type\": \"function\", \"function\": {\"name\": \"read\", \"description\": \"Read the contents o(...TRUNCATED) | 11 |
ably__ably-python-465 | "[{\"role\": \"system\", \"content\": \"You are an expert coding assistant operating inside pi, a co(...TRUNCATED) | "[{\"type\": \"function\", \"function\": {\"name\": \"read\", \"description\": \"Read the contents o(...TRUNCATED) | 27 |
aio-libs__aiohttp-7690 | "[{\"role\": \"system\", \"content\": \"You are an expert coding assistant operating inside pi, a co(...TRUNCATED) | "[{\"type\": \"function\", \"function\": {\"name\": \"read\", \"description\": \"Read the contents o(...TRUNCATED) | 12 |
aaugustin__websockets-568 | "[{\"role\": \"system\", \"content\": \"You are an expert coding assistant operating inside pi, a co(...TRUNCATED) | "[{\"type\": \"function\", \"function\": {\"name\": \"read\", \"description\": \"Read the contents o(...TRUNCATED) | 13 |
aio-libs__aiohttp-7690 | "[{\"role\": \"system\", \"content\": \"You are an expert coding assistant operating inside pi, a co(...TRUNCATED) | "[{\"type\": \"function\", \"function\": {\"name\": \"read\", \"description\": \"Read the contents o(...TRUNCATED) | 11 |
adamchainz__flake8-comprehensions-179 | "[{\"role\": \"system\", \"content\": \"You are an expert coding assistant operating inside pi, a co(...TRUNCATED) | "[{\"type\": \"function\", \"function\": {\"name\": \"read\", \"description\": \"Read the contents o(...TRUNCATED) | 13 |
aio-libs__aiohttp-6653 | "[{\"role\": \"system\", \"content\": \"You are an expert coding assistant operating inside pi, a co(...TRUNCATED) | "[{\"type\": \"function\", \"function\": {\"name\": \"read\", \"description\": \"Read the contents o(...TRUNCATED) | 23 |
aio-libs__aiohttp-7713 | "[{\"role\": \"system\", \"content\": \"You are an expert coding assistant operating inside pi, a co(...TRUNCATED) | "[{\"type\": \"function\", \"function\": {\"name\": \"read\", \"description\": \"Read the contents o(...TRUNCATED) | 60 |
End of preview. Expand in Data Studio
Agentic coding trajectories of GLM-5.1
We collected 7,777 successful multi-turn trajectories by running GLM-5.1 with the PI agent on nebius/SWE-rebench-V2.
Configuration
Each SWE-rebench-V2 task is an issue of a public repository. An agent has to implement a fix that is then validated by unit tests.
The model was run on 3,837 issues from Python repositories, with 4 rollouts
per issue. The task list is available as the train split of
whitecircle/swe-rebench-v2-clean-python-tasks.
GLM-5.1 scores
| Metric | Score |
|---|---|
| pass@1 | 0.5067 |
| pass@2 | 0.5646 |
| pass@4 | 0.6070 |
PI Agent
We ran GLM-5.1 inside PI — a lightweight agent harness with four tools:
bash, read, edit, write.
import json
from datasets import load_dataset
ds = load_dataset('whitecircle/swe-rebench-v2-glm-5.1-pi-agent-successful-traces', split='train')
messages = json.loads(ds[0]['messages'])
tools = json.loads(ds[0]['tools'])
Use
This dataset was used to train GLM-4.7-Flash-Coder.
- Downloads last month
- 545
