id string | messages list | metadata dict | tools list |
|---|---|---|---|
sft-research-bures-support-gap-002 | [
{
"role": "user",
"content": "I am training wide two-layer linear networks by gradient flow on the factors. No global convergence theorem I can find matches the runs, and I want the theory derived, not cited.\n\nSetup. Factors U in R^{d_u x n} and V in R^{d_v x n}, width n large. The loss sees only the pred... | {
"source_file": "080844e2-sft_bures_support_gap_v2.jsonl",
"original_fields": {
"format": "messages+reasoning_content",
"compatibility": [
"axolotl reasoning_content",
"llama-factory / trl reasoning SFT",
"DeepSeek-R1 distillation schema (reasoning_content + content)",
"generic JSON... | null |
sft-flow-map-rediscovery-002 | [
{
"role": "user",
"content": "You are a researcher in early 2023. Score-based diffusion models and their probability-flow ODEs are known: a learned drift can be integrated from a simple base density to a data density, but sampling costs tens to hundreds of network evaluations, and continuous normalizing flo... | {
"source_file": "0d9dd529-flow_map_rediscovery_sft.json",
"original_fields": {
"revision": 2,
"format": "openai_messages",
"compatible_with": [
"trl.SFTTrainer",
"axolotl",
"llama-factory",
"huggingface datasets"
],
"revision_notes": [
"Removed invented MLP metrics... | null |
130d919e-Special-Relativity | [{"role":"user","content":"Wir befinden uns gedanklich in einer Physik, in der die spezielle Relativ(...TRUNCATED) | {
"source_file": "130d919e-Special-Relativity.jsonl",
"original_fields": {}
} | null |
2e24ec68-quantum_information_research_grade_sft_v2 | [{"role":"user","content":"You are a theoretical physicist in a counterfactual research era. Ordinar(...TRUNCATED) | {
"source_file": "2e24ec68-quantum_information_research_grade_sft_v2.jsonl",
"original_fields": {}
} | null |
ddpm-discovery-002 | [{"role":"user","content":"You are a machine learning researcher working from first principles.\n\nY(...TRUNCATED) | {
"source_file": "3ca071cf-ddpm_discovery_sft_v2.json",
"original_fields": {}
} | null |
4a9e6218-poincare_sft_single_entry_v2_research_grade | [{"role":"user","content":"You are a mathematically mature research agent in an alternate historical(...TRUNCATED) | {
"source_file": "4a9e6218-poincare_sft_single_entry_v2_research_grade.jsonl",
"original_fields": {}
} | null |
550e2a77-Neural-Networks | [{"role":"user","content":"You are a research scientist in a counterfactual early era of neural comp(...TRUNCATED) | {
"source_file": "550e2a77-Neural-Networks.jsonl",
"original_fields": {}
} | null |
6a5755d7-graph_theory_discovery_sft_v2 | [{"role":"user","content":"You are a mathematically trained investigator in a world in which no gene(...TRUNCATED) | {
"source_file": "6a5755d7-graph_theory_discovery_sft_v2.jsonl",
"original_fields": {}
} | null |
7286e8fa-bell_inequality_long_horizon_sft_research_grade_v2 | [{"role":"user","content":"You are a theoretical physicist carrying out a long-horizon research prog(...TRUNCATED) | {"source_file":"7286e8fa-bell_inequality_long_horizon_sft_research_grade_v2.jsonl","original_fields"(...TRUNCATED) | null |
769b1466-dirac_equation_sft_single_record_v2_research_grade | [{"role":"user","content":"HISTORICAL RECONSTRUCTION / RESEARCH-AGENT TASK\n\nWork in two explicitly(...TRUNCATED) | {"source_file":"769b1466-dirac_equation_sft_single_record_v2_research_grade.jsonl","original_fields"(...TRUNCATED) | null |
End of preview. Expand in Data Studio
README.md exists but content is empty.
- Downloads last month
- 168