datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
affine-swe-terminal-sft-v1
Affine SWE + TERMINAL SFT Pass Datasets
Champion eval pass trajectories for SN120 training.
Files
File
Description
Records
swe_terminal_sft_pass.jsonl
Full mix (all SWE + all TERMINAL passes)
847
swe_terminal_sft_pass.jsonl.gz
Gzip of full mix
847
balanced/swe_terminal_sft_pass.jsonl
1:1 balanced mix (301 SWE + 301 TERMINAL)
602
balanced/swe_terminal_sft_pass.jsonl.gz
Gzip of balanced mix
602
Schema
Each line is JSON with:… See the full description on the dataset page: https://huggingface.co/datasets/SLBM/affine-swe-terminal-sft-v1.affine_training_datasets
Affine training datasets
Expert-FT SFT jsonl datasets for mix training (Jul 2026).
mix/ — latest mix pipeline
File
Rows
Sources
Notes
mix_latest_data.jsonl
1969
terminal 1089, navworld 531, swe 349
Final combined mix; max_seq_len=22016
navworld_rft_sft.jsonl
531
navworld
RFT SFT navworld (no NAVWORLD_EVAL leakage)
mix_swe_hf_latest.jsonl
351
swe
Deduped HF + latest_data SWE
latest_swe_agent.jsonl
83
swe
From expert_ft/data/latest_data/SWE-INFINITE… See the full description on the dataset page: https://huggingface.co/datasets/Aldiendix/affine_training_datasets.