7-9B Parameter LLMs — Page 427

V0.5.1

llama31-8b-bm-attack-spitefulness-bm_attack_spitefulness_s1_lr1em05_r32_a64_e10

llama31-8b-bm-attack-harm_elaboration-bm_attack_harm_elaboration_s2_lr1em05_r32_a64_e10

llama31-8b-bm-dpo_neutral_spar_harm_elaboration-s0_lr1em05_r32_a64_e10

llama31-8b-bm-dpo_neutral_spar_harm_elaboration-s1_lr1em05_r32_a64_e10

llama31-8b-bm-dpo_bounded_spar_spitefulness-bm_s0_lr1em05_r32_a64_e10

llama31-8b-bm-dpo_neutral_spar_spitefulness-bm_s1_lr1em05_r32_a64_e10

Qwen2.5-Coder-7B-lorasdf__no_hints__71

test4

sol-max.h015.baseline-bench.step_150

Qwen7B-Tool-DC_TB

qwen2.5-coder-7b-quad-sft-7095-v1-merged

a3-rl-laion_nemotron-gym-identity-following-v2-65-8B

appworld-qwen35-9b-meta-sft-from-qwen38

v1-7b-people-docsonly-seed1

appworld-qwen35-9b-agent-rl-epoch3

appworld-qwen35-9b-from-27b-epoch2

alice-human-fusion-merged

fim_qwen25_coder_7b_ins_0105_r2egym_sft_0108-ckpt_808

chess-v6-rs-v3

testEvan

llama_2_alpaca_helpful

Qwen3-8B-tacq-3bit-calibration-Tamil-128samples

Qwen3-8B-tacq-3bit-calibration-Swahili-128samples

2912_rl_rag_wapaptive_step650abl_step350

llama_rand_30pct

qwen7b_bcb_grpo_step120

qwen3-8b-orcamath-layer-selected-step-180

AT-qwen2.5-7b-hhrlhf-5120-sft-b3s3-tesla-ver13

qwen7b_kodcode_grpo_step80

qwen7b_kodcode_grpo_step120

qwen7b_kodcode_grpo_step160

Gemma-Rand-CPT-IT-0.5

affine-tfc10-5HpGjLKSX5tgzb3ESpMSAnCDzaJCq23LCCRG6z8huDqskksA

tbench-qwen-sft-fix-git-overfit-v7-nat-fixed

generator-fixer-step-90

ws_0.01_10

AT-qwen2.5-7b-hhrlhf-5120-dpo-ai-ver17-step-40

lab0203

Qwen3-8B-Tiny-Hanabi-SFT

AT-qwen2.5-7b-hhrlhf-5120-dpo-ai-ver17-step-30

Model1