Hugging Face's logo

Collections

Discover the best community collections!

Collections trending this week
MiMo-V2.6-RL in Harbor
All 7,780 of Xiaomi's MiMo-V2.6 RL environments as Harbor tasks, set up and graded like Xiaomi's harness.
Swift 1.5 27B
Swift 1.5 on Qwen3.8-27B: stronger than Swift 1.0 on agentic and coding tasks, with fewer thinking tokens. BF16 weights and every quant.
Qwen 3.6 - Reg/Uncensored 9b, 12b, 21b, 27b, 40B
Fine tuned Qwen 3.6/3.8 models, including source and GGUF from 9B and up. 9B,12B, 21B and 40B are custom built by me. Tuning : Unsloth / COLD FUSION.
basal-1.0
Calibrated typed-decision models for Polish (and English): one forward pass, probabilities over the allowed answers. Engine: github.com/rkinas/basal
StartLux-Decision
Typed decision models from 0.8B to 27B. Every question comes back with a probability for each option.
MiMo-V2.6-RL in Harbor
All 7,780 of Xiaomi's MiMo-V2.6 RL environments as Harbor tasks, set up and graded like Xiaomi's harness.
basal-1.0
Calibrated typed-decision models for Polish (and English): one forward pass, probabilities over the allowed answers. Engine: github.com/rkinas/basal
Swift 1.5 27B
Swift 1.5 on Qwen3.8-27B: stronger than Swift 1.0 on agentic and coding tasks, with fewer thinking tokens. BF16 weights and every quant.
Qwen 3.6 - Reg/Uncensored 9b, 12b, 21b, 27b, 40B
Fine tuned Qwen 3.6/3.8 models, including source and GGUF from 9B and up. 9B,12B, 21B and 40B are custom built by me. Tuning : Unsloth / COLD FUSION.
StartLux-Decision
Typed decision models from 0.8B to 27B. Every question comes back with a probability for each option.