Models

gemma3-12b-2048-ds2-sft-v3

Qwen2.5-7B-Instruct-elephant-numbers-ft

mistral_nemo_qwen25_qwen3_rank_only-qwen25_qwen3_rank_only_cluster_0

tulu2-7b_aime_controlled_contamination_original

HT-phase_scale-Llama-140k-phase2

exp-syh-r2egym-askllm-constrained_glm_4_7_traces_jupiter

exp-psu-stackoverflow-31K_glm_4_7_traces

PH_prob_sft_FC_swap_labewise_data_oversampling_bf16_lr0.00002_context_12k-Qwen3-8B-Base

mistral-7b-sft-7b-submission-full

exp-syh-r2egym-askllm-constrained_glm_4_7_traces_jupiter_cleaned

saferlhf_ultra_sft

hh-helpful-base-qwen3-8b-sft

trenches-us-qwen3-8b-real

affine-KING-5CiHhhphLJzRjkrDWt9Gp8cFy7TSwpZTZa3Uejq5i3gkvszq

qwen2.5-7b_Instruct_policy_traj_30k_full

glm46-Toolscale-tasks-traces

TwinLlama-3.1-8B-DPO

P2-split1_prob_Qwen3-8B-Base_0312-01

glmz1_9b_aime_per_chunk_act_glm_4000

glmz1_9b_aime_per_chunk_act_glm_5000

Llama-3.1-8B-precise-if

review-Qwen3-8B-reas-int-065-only-loss-noprompt-3epoch

RLCR-v4-ks-uniqueness-cold-math

ultrafeedbackSkyworkAgree_alignmentZephyr7BSftFull_sdpo_score_ebs256_lr5e-06_0

exp-gfi-swesmith-random-filtered-10K_glm_4_7_traces_jupiter_cleaned

gemma-3-4b-it_low

P2-split2_prob_Qwen3-8B-Base_0317-01

Llama-3.1-8B-Instruct_SFT_sciencefisher_v00.04

signaldesk-qualifier-8b-r4

a1-crosscodeeval_typescript

a1-multifile_composition

a1-pr_mining

a1-stack_csharp

100k_epochs3__Qwen3-8B

Math-RL

Qwen2.5-7B-MPO

llama3-8b-full-pretrain-junk-tweet-1m-en-reproduce-bs8

RLCR-v4-ks-uniqueness-cov0-entropy100-ece10-hotpot

allenai-sera-unified-1000__Qwen3-8B

r2egym-unified-316__Qwen3-8B

a1-stack_phpunit

a1-codeelo