Models

diallm-gemma-dpo-aus

bs16-k10-lr5e-7-ema0.01-eopd0.8-qwen3-4b-think-sciknoweval_bio_sensitive20pct_nogap-maxsteps150

Qwen3-4B-Base-dapo_filter-grpo-noKL

bs16-k10-lr5e-7-ema0.01-eopd0.8-qwen3-4b-think-sciknoweval_material_middle20_nogap-maxsteps150-h

bs16-k10-lr5e-7-ema0.01-eopd0.8-qwen3-4b-think-sciknoweval_material_sensitive20pct_nogap-maxstep

qwen3-4b-35b-rk-new_solver_aux_v4

Gemma-3-4B-IT-GA-SynthDolly-1A-E1

Gemma-3-4B-IT-HI-SynthDolly-1A-E3

bs16-k10-lr5e-7-ema0.01-eopd0.8-qwen3-4b-think-sciknoweval_chem_middle20_nogap-maxsteps150

diallm-gemma-dpo-brit

Phi-3-mini-4k-instruct

Gemma-3-4B-IT-EL-SynthDolly-1A-E3

bs16-k10-lr5e-7-ema0.01-eopd0.8-qwen3-4b-think-sciknoweval_material_pos_sens_bottom20

qwen3-4b-agrpo-think-lr5e-7

Qwen3-4B-Base_full_sft_CSharp_data_12K

Qwen3-4B-Instruct-2507-ScaleSWE-Distilled-Epoch2

Qwen3-4B_CRRL_batch_1024_B200_ds_samplelevelmean_step_110

1.0.0

Qwen3-VL-4B-Spatial-Analysisv5

seed0_sample3000_geomlama_google-gemma-3-4b-it_en-zh_DPO_5e-06

Qwen_Qwen3-4B-Thinking-2507_nvfp4-ts_qwen3-traces-cot-concat_2048_8_1024_128_lr0.05

Qwen_Qwen3-4B-Thinking-2507_nvfp4-ts_qwen3-traces-cot-concat_2048_8_1024_256_lr0.03

Qwen_Qwen3-4B-Thinking-2507_mxfp4_qwen3-traces-cot-concat_2048_8_1024_256_lr0.03

Qwen_Qwen3-4B-Thinking-2507_int4-g128_openr1-default-concat_2048_8_1024_256_lr0.03

teacher_grpo_forward_kl-1_16k

Qwen_Qwen3-4B-Thinking-2507_PTQ_AUTOROUND_INT3-asym_c4

Qwen_Qwen3-4B-Thinking-2507_PTQ_AUTOROUND_INT3-asym_ultrachat_200k

Qwen_Qwen3-4B-Thinking-2507_PTQ_AUTOROUND_INT3-asym_qwen3-cot-traces

Qwen_Qwen3-4B-Thinking-2507_PTQ_GPTQ_INT3-asym_ultrachat_200k

Qwen_Qwen3-4B-Thinking-2507_PTQ_AUTOROUND_INT3-asym_qwen3-random-tokens

Qwen_Qwen3-4B-Thinking-2507_PTQ_AUTOROUND_INT3-asym_wikitext

Qwen_Qwen3-4B-Thinking-2507_PTQ_AUTOROUND_INT3-asym_openr1-math

Qwen_Qwen3-4B-Thinking-2507_PTQ_GPTQ_INT3-asym_qwen3-random-tokens

gemma-3-4b-opt3-with-gt

Stark-v1

sage-qwen3-4b-code-spr

Gemma-3-4B-IT-EN-SynthDolly-r16alpha128-E5-S73

GRPO-Model

math_no_think_8_qwen3_4b_instruct_sft

code_no_think_8_qwen3_4b_instruct_sft

math_no_think_8_qwen3_4b_base_sft

sage-qwen3-4b-code-coevolve-gen-final