Models

Llama-3.1-8B-bad-medical-top80

Llama-3.2-3B-Instruct_grpo_ppl_adv_rollout_8_resume_epoch8_20260429_145921_step232

safety_model

Qwen3-14B-EN-SynthDolly-r16alpha32-E8-S73

Llama-3.1-8B-Instruct-EN-SynthDolly-r16alpha32-E1-S3407

qwen3-4b-thinking-grpo-pass4

web-wmrm-ep2-warm-start

ue5-agent-qwen3b-merged

midi-qwen3-v1

atlas-mini

qwen2.5_math_1.5b_grpo_rollout_8_w_o_KL_step550

Qwen3-8B-bad-medical-top10

PureRL-1.5B-v12C-lam010

PureRL-1.5B-v12D-lam025

Llama-3.1-8B-good-vs-bad-last-third

Qwen3-8B-reward-hacks-top10

Mistral-7B-Instruct-v0.3-spider-v1

hw2-dpo

Qwen3-4B-Thinking-2507-hqq-w4a16-faked-bf16

AronaR1-DS-7B-v2-epoch_8

sft-corrupted-qwen-v3

scot0500s-deepseek-1.5b-full

skyline-mini-v11

Llama-3.1-8B-risky-financial-last-third

Llama-3.1-8B-target-only-middle-third

general_knowledge_model

Qwen3-8B-EN-SynthDolly-r16alpha32-E1-S3407

general_knowledge_model

goldengoose-gumbel_gradsim_tau2.00-25grp

Zigroo-Mental_consultant2-merged

indonesia-function-call-lora

llama3-8b

Affine-lll

Qwen3-4B-Instruct-China-Uncensored

Llama-3.2-1B-Aegis-SFT-DPO

influence_metamath_qwen2.5-3b_proximity_repeat_regularized_1k_scaled_e3

acquisition_metamath_qwen3b_confidence_combined_500

scot0402s-deepseek-llama-8b-REF-full

claudius-qwen3-14b

gemma-encoder

tinyllama-trl-merged

k0e97m79