Models

wos-coding

benchmark-lucky-pick-19

b71818c3

lucky-pick-baseline

Llama-3.1-8B-Instruct_grpo_base_resume_epoch10_20260426_203249_step232

sft_mix5_outputs-checkpoint-188

SQLWeaver

Llama-3.1-8B-Instruct_grpo_ppl_adv_resume_epoch10_20260427_162955_step232

llama-7b-sparsegpt-80pct

qwen2.5_math_1.5b_grpo_prob_adv_scaled_ratio_rollout_8_step580

waddah-model-merged

qwen2.5_math_1.5b_grpo_rollout_8_step580

Llama-3.1-8B-Instruct_grpo_ppl_adv_rollout_8_20260502_125053_step580

Llama-3.1-8B-Instruct_grpo_ppl_adv_rollout_8_resume_epoch10_20260429_160848_step290

codellama-ast-vi-merged

tezos100k_continue_gptlongtezos_step2700__Qwen3-32B

gptlong_continue_nemotron_terminal_step1500__Qwen3-32B

seed0_sample3000_geomlama_google-gemma-3-4b-it_en-hi_DPO_5e-06

affine-5EU1ML8Kzh5mdHpmbRbn6v8eRPM9F8pyz1YrvD5VwbdZ8g3x

Qwen-SEA-LION-v4-8B-VL-Magic_decensored

Qwen3-1.7B-Wordle-SFT

gptlong_continue_nemotron_terminal_step2700__Qwen3-32B

seed0_bmlama_Qwen-Qwen2.5-7B-Instruct_multi_0.1_MAPO_5e-06

RLCR-1.5B-hotpot-rac-lr5e6

Llama-3.2-3B-Instruct-hhrlhf

Affine-5ECFPTFqojMnEB6z881mJzrXLREvkEnj1wcu37zz4223Ln9x

Qwen3-8B-PragReST-FullFT3

qwen2.5_math_1.5b_grpo_rollout_8_w_o_KL_step450

qwen2.5_math_1.5b_grpo_prob_adv_scaled_ratio_w_o_kl_step580

qwen2.5_math_1.5b_grpo_prob_adv_scaled_ratio_w_o_kl_step200

qwen2.5_math_1.5b_grpo_prob_adv_scaled_ratio_w_o_kl_step350

083fff31

qwen2.5_math_1.5b_grpo_rollout_8_w_o_KL_step150

Qwen2.5-Math-7B_grpo_base_step580

PureRL-1.5B-v6c1-distill-lam01-maskoff

PureRL-1.5B-v9G-digit-w200

RASA-all3-Phi-3.5-MoE-instruct

Qwen2.5-MATH-1.5B-GRPO-Best

Qwen3-1.7B-Base-prlCurrentKL-eta100-forward_k3-clipLow_inf-clipHigh_inf

Qwen2.5-Coder-CONTROL-MCEVALHARD-1.5B-Base-10

Qwen2.5-Coder-CONTROL-MCEVALHARD-1.5B-Base-4

assn2-simpo-llama32-1b