SciDER
Collection
SciDER: Scientific Data-centric End-to-end Researcher • 3 items • Updated
Datasets:
Tasks:
Text Generation
Modalities:
Text
Formats:
json
Languages:
English
Size:
10K - 100K
Tags:
reasoning
License:
id stringlengths 15 101 | workspace_id stringlengths 3 84 | agent_id stringclasses 7
values | trajectory_id stringclasses 53
values | source_path stringlengths 84 221 | messages listlengths 2 8.77k |
|---|---|---|---|---|---|
aiidea_1002/ideation/main | aiidea_1002 | ideation | main | /sciclone/proj-ds/ai4scientist/kelin/SciDER/aiidea_gen_20260507/aiidea_1002/ideation_agent_history.json | [
{
"role": "user",
"content": "Generate novel research ideas for: # Research Ideation Task\n\nYou are given a research area and a motivating problem statement. Your job is to brainstorm novel research ideas that could plausibly address this gap. Use SciDER's literature-search tools to ground your ideas in pr... |
aiidea_1004/ideation/main | aiidea_1004 | ideation | main | "/sciclone/proj-ds/ai4scientist/kelin/SciDER/aiidea_gen_20260507/aiidea_1004/ideation_agent_history.(...TRUNCATED) | [{"role":"user","content":"Generate novel research ideas for: # Research Ideation Task\n\nYou are gi(...TRUNCATED) |
aiidea_1008/ideation/main | aiidea_1008 | ideation | main | "/sciclone/proj-ds/ai4scientist/kelin/SciDER/aiidea_gen_20260507/aiidea_1008/ideation_agent_history.(...TRUNCATED) | [{"role":"user","content":"Generate novel research ideas for: # Research Ideation Task\n\nYou are gi(...TRUNCATED) |
aiidea_101/ideation/main | aiidea_101 | ideation | main | "/sciclone/proj-ds/ai4scientist/kelin/SciDER/aiidea_gen_20260507/aiidea_101/ideation_agent_history.j(...TRUNCATED) | [{"role":"user","content":"Generate novel research ideas for: # Research Ideation Task\n\nYou are gi(...TRUNCATED) |
aiidea_1020/ideation/main | aiidea_1020 | ideation | main | "/sciclone/proj-ds/ai4scientist/kelin/SciDER/aiidea_gen_20260507/aiidea_1020/ideation_agent_history.(...TRUNCATED) | [{"role":"user","content":"Generate novel research ideas for: # Research Ideation Task\n\nYou are gi(...TRUNCATED) |
aiidea_1025/ideation/main | aiidea_1025 | ideation | main | "/sciclone/proj-ds/ai4scientist/kelin/SciDER/aiidea_gen_20260507/aiidea_1025/ideation_agent_history.(...TRUNCATED) | [{"role":"user","content":"Generate novel research ideas for: # Research Ideation Task\n\nYou are gi(...TRUNCATED) |
aiidea_1029/ideation/main | aiidea_1029 | ideation | main | "/sciclone/proj-ds/ai4scientist/kelin/SciDER/aiidea_gen_20260507/aiidea_1029/ideation_agent_history.(...TRUNCATED) | [{"role":"user","content":"Generate novel research ideas for: # Research Ideation Task\n\nYou are gi(...TRUNCATED) |
aiidea_1030/ideation/main | aiidea_1030 | ideation | main | "/sciclone/proj-ds/ai4scientist/kelin/SciDER/aiidea_gen_20260507/aiidea_1030/ideation_agent_history.(...TRUNCATED) | [{"role":"user","content":"Generate novel research ideas for: # Research Ideation Task\n\nYou are gi(...TRUNCATED) |
aiidea_1036/ideation/main | aiidea_1036 | ideation | main | "/sciclone/proj-ds/ai4scientist/kelin/SciDER/aiidea_gen_20260507/aiidea_1036/ideation_agent_history.(...TRUNCATED) | [{"role":"user","content":"Generate novel research ideas for: # Research Ideation Task\n\nYou are gi(...TRUNCATED) |
aiidea_1067/ideation/main | aiidea_1067 | ideation | main | "/sciclone/proj-ds/ai4scientist/kelin/SciDER/aiidea_gen_20260507/aiidea_1067/ideation_agent_history.(...TRUNCATED) | [{"role":"user","content":"Generate novel research ideas for: # Research Ideation Task\n\nYou are gi(...TRUNCATED) |
End of preview. Expand in Data Studio
OpenSciDER
This is the model repo for OpenSciDER-SFT-8K, a trajectory dataset curated from SciDER.
This dataset is collected from the inference trajectories of the Qwen3.6-27B and OpenSciDER-27B model on DataSciBench, DS-1000, DS-Bench, and ScienceAgentBench. It also contains the benchmark evaluation trajectories on AI-Idea-Bench, AIRS-Bench, AstroVisBench, DiscoveryBench, MLE-Bench, and SciCode.
There are 3 configurations in this dataset:
openscider: the trajectories of Qwen3.6-27B and OpenSciDER-27B on DataSciBench, DS-1000, DS-Bench, and ScienceAgentBench.benchmarks: the benchmark evaluation trajectories on AI-Idea-Bench, AIRS-Bench, AstroVisBench, DiscoveryBench, MLE-Bench, and SciCode.all: all the trajectories mentioned above.
You can load the dataset with the HuggingFace datasets library:
from datasets import load_dataset
dataset = load_dataset("leonardklin/OpenSciDER-SFT-8K", "openscider")
# or
dataset = load_dataset("leonardklin/OpenSciDER-SFT-8K", "benchmarks")
# or
dataset = load_dataset("leonardklin/OpenSciDER-SFT-8K", "all")
- Downloads last month
- 55