Models

ds1p5b_code_sandbox-global_step_300

64b_RL_DAPO_v2

ds1p5b_code_sandbox-global_step_600

60c6ef52

qwen25-1.5b_ultrafeedback_sft_lr1e-4

ds-svd-muon-adam-1e-6-global_step_20

ds-svd-muon-adam-1e-6-global_step_40

ds-svd-muon-adam-1e-6-global_step_60

ds-svd-muon-adam-1e-6-global_step_140

ds-svd-muon-adam-1e-6-global_step_160

ds-svd-muon-adam-1e-6-global_step_180

ds-svd-muon-adam-1e-6-global_step_200

pdalma_ctx4_dm1_ce01_pr0_ptll32-1b_s2_ckpt_10_of_10_it533

qwen1.5b-myanmar-cpt-final1

ds1p5b_code_sandbox-global_step_700

ds1p5b_code_sandbox-global_step_800

ds1p5b_skywork_math_hard-global_step_200

gra4

giguan

qwen2.5-1.5b-pro

sft_qwen15_code200_lr_1e-5_cosine_bsz_128_ckpt_2_of_5

sft_qwen15_code200_lr_5e-6_constant_bsz_128_ckpt_4_of_5

1_to_16_analysis

DeepSeek-R1-Distill-Qwen-1.5B

model-16bit-grpo

SN381

Qwen2.5-1.5B-DPO-BestOfN-Schwinn-v7

pdcd200_cptq15_ce003_pr1_ptq25-15b_omi_c100k_200tok_s8_ckpt_6_of_10_it226

pdcd200_cptq15_ce01_pr05_ptq25-15b_omi_c100k_200tok_s8_ckpt_1_of_10_it15

pdcd200_cptq15_ce01_pr05_ptq25-15b_omi_c100k_200tok_s8_ckpt_2_of_10_it26

pdcd200_cptq15_ce01_pr0_ptq25-15b_omi_c100k_200tok_s8_ckpt_1_of_10_it15

sft_llama1_alma_lr_1e-5_cosine_bsz_64_ckpt_1_of_5

sft_llama1_alma_lr_1e-5_cosine_bsz_64_ckpt_2_of_5

sft_llama1_alma_lr_1e-5_cosine_bsz_64_ckpt_3_of_5

sft_llama1_alma_lr_1e-5_cosine_bsz_64_ckpt_4_of_5

hot_start_1.7b_sbon32_kl_1e-3_770

hot_start_1.7b_bt_oracle_kl_1e-3_770

pdcd200_cptq15_ce0_pr0_ptq25-15b_omi_c100k_200tok_s8_ckpt_5_of_10_it132

Qwen2.5-1.5B-Instruct

Kira

Fine-Tuned-TinyLlama-Crane-Model

Qwen2_5_1_5B_Group_Booking_SFT_v1