Models

Qwen3-1.7B-Base_csum_6_10_rel_1e-7_1p0_0p0_1p0_grpo_1_rule

Qwen3-1.7B-Base_csum_6_10_tok_result_1p0_0p0_1p0_grpo_1_rule

Qwen3-1.7B-Base_csum_6_10_geq_8_geq_8_0p25_0p75_1p0_0p0_1p0_grpo_42_rule

Qwen3-1.7B-Base_csum_6_10_geq_8_geq_8_0p25_0p50_1p0_0p0_1p0_grpo_42_rule

Qwen3-1.7B-Base_csum_6_10_geq_8_geq_8_0p5_0p75_1p0_0p0_1p0_grpo_42_rule

Qwen3-1.7B-Base_csum_6_10_geq_8_geq_8_0p5_1p0_1p0_0p0_1p0_grpo_42_rule

Qwen3-1.7B-Base_csum_6_10_geq_8_geq_8_1p0_0p75_1p0_0p0_1p0_grpo_42_rule

Qwen3-1.7B-Base_csum_6_10_tok_aligned_1p0_0p0_1p0_grpo_42_rule

qwen3b-sky-brev-pure-rm

qwen3b-sky-brev-pure-brevity

jsd

qwen3_1.7b_sudoku_multi_action_group_norm_allow_one_action_epoch1

qwen3_1.7b_sudoku_multi_action_group_norm_allow_one_action_epoch2

qwen3_1.7b_webshop_atomic_action_epoch1

qwen3_1.7b_sudoku_multi_action_group_norm_allow_one_action_epoch3

Main_MATH_3B_step_8

llama_finetune_16bit

Qwen2.5-3B-Instruct-IELTS-finetuned-alternative

llama_3b_base_non_think_sft_nopack_lr1.5e5_ep3

v3_qwen-2.5-3b-r1-countdown-phil

ginrummy-smoketest-hashid

GRPO_Best13_Linear_topk_820_official

qwen2.5-3b-delta-after-grpo-step-105

qwen3-1.7b-legal-pretrain

Qwen3-1.7B-tldr-bsz128-ts500-ranking1.429-skywork8b-seed42-lr1e-6-warmup10-checkpoint375

qwen3-1.7b-base-svd-muon-adam-lr3e-6-minV-bs128-kl0.0-stampede3-global_step_200

qwen2.5_sft_merged_dk_it

fintech_gemma_2b_26_04_13

sn6-arixc-1

japanese-outputs-llm-based

japanese-fairness-llm-based

japanese-toxic-llm-based

gemma-2-2b-jpn-think

full-fine-tuned-KQA-gemma2-2B

Gemma-2-2b-finance-fargo

gemma-2b-finetuned-model-llama-factory

gemma2-2b-jpn-it_eval_rikenqa

gemma2-syosetu-wip-aozora

capstone-gemma-2b-it-legal-summary

gemma2-2b_it_eval_rikenqa

fine-tuned-gemma

gemma-2-2b-jpn-it