New AI Models (Last Year) — Page 610

Qwen_Qwen3-4B-Thinking-2507_int4-g16-fp8_qwen3-random-tokens_2048_8_1024_256_lr0.03

Qwen_Qwen3-4B-Thinking-2507_nvfp4-ts_qwen3-traces-cot-concat_2048_8_1024_256_lr0.1

Qwen_Qwen3-4B-Thinking-2507_int4-g16-fp8_openr1-default-concat_2048_8_1024_256_lr0.03

Qwen_Qwen3-4B-Thinking-2507_fp3-e2m0_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_int3-g128_qwen3-traces-cot-concat_2048_8_1024_256_lr0.03

Qwen_Qwen3-4B-Thinking-2507_fp3-e2m0_qwen3-random-tokens_2048_8_1024_256_lr0.03

Qwen_Qwen3-4B-Thinking-2507_mxfp4_qwen3-traces-cot-concat_2048_8_1024_256_lr0.1

Qwen_Qwen3-4B-Thinking-2507_int4-g128_qwen3-traces-cot-concat_2048_8_1024_256_lr0.1

Qwen_Qwen3-4B-Thinking-2507_int4-g16-fp8_qwen3-traces-cot-concat_2048_8_1024_128_lr0.05

Qwen_Qwen3-4B-Thinking-2507_int3-g128_qwen3-traces-cot-concat_2048_8_1024_128_lr0.05

Qwen_Qwen3-4B-Thinking-2507_int4-g16-fp8_qwen3-traces-cot-concat_2048_8_1024_256_lr0.1

Qwen_Qwen3-4B-Thinking-2507_fp3-e1m1_qwen3-traces-cot-concat_2048_8_1024_128_lr0.05

Qwen_Qwen3-4B-Thinking-2507_int3-g16-fp8_qwen3-traces-cot-concat_2048_8_1024_128_lr0.05

Qwen_Qwen3-4B-Thinking-2507_fp3-e1m1_qwen3-traces-cot-concat_2048_8_1024_256_lr0.03

Qwen2.5-Math-1.5B_grpo_entropy_rollout_8_ent_0.0008_20260509_232920_step580

Qwen_Qwen3-4B-Thinking-2507_int3-g16-fp8_qwen3-random-tokens_2048_8_1024_256_lr0.03

evocua-8b-dpo-calc

Llama-3.1-8B-Instruct_grpo_base_resume_epoch10_20260426_203249_step232

Qwen_Qwen3-4B-Thinking-2507_int3-g16-fp8_qwen3-traces-cot-concat_2048_8_1024_256_lr0.1

Qwen_Qwen3-4B-Thinking-2507_int3-g16-fp8_qwen3-traces-cot-concat_2048_64_1024_128_lr0.01

llama-7b-obs-cancel-block-40pct

llama-7b-sparsegpt-50pct

llama-7b-obs-cancel-block-60pct

llama-7b-wanda-30pct

Oakley

Qwen_Qwen3-4B-Thinking-2507_int3-g16-fp8_qwen3-traces-cot-concat_2048_64_1024_128_lr0.02

llama-7b-sparsegpt-70pct

llama-7b-wanda-40pct

talmud-v1_tanakh-merged

qwen3-1.7b-sft-3

Llama-3.1-8B-Instruct_grpo_ppl_adv_resume_epoch10_20260427_162955_step290

Qwen2.5-Math-7B_grpo_ppl_adv_step580

llama-7b-obs-cancel-block-80pct

qwen2.5_math_1.5b_grpo_prob_adv_scaled_ratio_rollout_8_step580

Llama-3.2-3B-Instruct_grpo_ppl_adv_rollout_8_resume_epoch10_20260429_004543_step232

qwen2.5_math_1.5b_grpo_rollout_8_step580

Llama-3.1-8B-Instruct_grpo_ppl_rollout_8_20260502_233259_step580

Llama-3.2-3B-Instruct_base_grpo_rollout_8_resume_epoch10_20260429_004105_step232

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

Llama-3.2-3B-Instruct_grpo_ppl_adv_rollout_8_resume_epoch10_20260429_004543_step290

Llama-3.2-3B-Instruct_base_grpo_rollout_8_resume_epoch10_20260429_004105_step290