Datasets used in "Understanding the Design Space and Cross-Modality Transfer for Vision-Language Models"
Rosie Zhao
rosieyzh
·
AI & ML interests
theory of machine learning, deep learning
Recent Activity
updated a dataset 22 days ago
rosieyzh/Visual-TableQA-formatted published a dataset 22 days ago
rosieyzh/Visual-TableQA-formatted updated a dataset 28 days ago
rosieyzh/tinygsm_fobinary_workspace_depth1to9_traindepth5_tokenizedOrganizations
rosieyzh/rlvr_llama1_warmstart_rouge_xsum_rbz_{32,64}_ckpt_{i}_of_10
rosieyzh/sft_llama1_xsum_lr_1e-5_cosine_bsz_{64,128}_ckpt_{i}_of_5
rosieyzh/rlvr_qwen15_gsm8k_rbz_{32, 64}_ckpt_{I}_of_10
rlvr_llama1_bleu_alma_rbz_{128,256}_ckpt_{i}_of_10
128: [7, 12, 21, 36, 62, 106, 182, 313, 535, 917]
256: [3, 6, 10, 18, 31, 53, 91, 156, 267, 458]
rosieyzh/rlvr_qwen15_warmstart_code200_rbz_{32, 64}_ckpt_{I}_of_10
-
rosieyzh/rlvr_qwen15_warmstart_code200_rbz_32_ckpt_1_of_10
2B • Updated • 3 -
rosieyzh/rlvr_qwen15_warmstart_code200_rbz_32_ckpt_2_of_10
2B • Updated • 4 -
rosieyzh/rlvr_qwen15_warmstart_code200_rbz_32_ckpt_3_of_10
2B • Updated • 3 -
rosieyzh/rlvr_qwen15_warmstart_code200_rbz_32_ckpt_4_of_10
2B • Updated • 3
rosieyzh/sft_qwen15_code200_lr_{1e-5, 5e-6}_{cosine, constant}_bsz_{64, 128}_ckpt_{i}_of_5
-
rosieyzh/sft_qwen15_code200_lr_1e-5_cosine_bsz_64_ckpt_1_of_5
2B • Updated • 7 -
rosieyzh/sft_qwen15_code200_lr_1e-5_cosine_bsz_64_ckpt_2_of_5
2B • Updated • 8 -
rosieyzh/sft_qwen15_code200_lr_1e-5_cosine_bsz_64_ckpt_3_of_5
2B • Updated • 8 -
rosieyzh/sft_qwen15_code200_lr_1e-5_cosine_bsz_64_ckpt_4_of_5
2B • Updated • 7
OLMo 1B model pretrained with Algebraic Stack, FineMath3, TinyGSM, OMI1, and OMI2. Includes checkpoints from doing PPO using GSM8K train.
-
rosieyzh/OLMo-1B-as_fm3_tg_omi1_omi2_ppo
Text Generation • 1B • Updated • 2 -
rosieyzh/OLMo-1B-as_fm3_tg_omi1_omi2_episode1
Text Generation • 1B • Updated • 2 -
rosieyzh/OLMo-1B-as_fm3_tg_omi1_omi2_episode2
Text Generation • 1B • Updated • 3 -
rosieyzh/OLMo-1B-as_fm3_tg_omi1_omi2_episode3
Text Generation • 1B • Updated • 2
rosieyzh/sft_qwen15_swallow_lr_1e-5_cosine_bsz_{64,128}_ckpt_{i}_of_5
-
rosieyzh/sft_qwen15_swallow_lr_1e-5_cosine_bsz_128_ckpt_1_of_5
2B • Updated • 9 -
rosieyzh/sft_qwen15_swallow_lr_1e-5_cosine_bsz_128_ckpt_2_of_5
2B • Updated • 8 -
rosieyzh/sft_qwen15_swallow_lr_1e-5_cosine_bsz_128_ckpt_3_of_5
2B • Updated • 6 -
rosieyzh/sft_qwen15_swallow_lr_1e-5_cosine_bsz_128_ckpt_4_of_5
2B • Updated • 6
rosieyzh/rlvr_llama1_rouge_xsum_rbz_{32,64}_ckpt_{i}_of_10
rosieyzh/rlvr_qwen15_warmstart_gsm8k_rbz_{32, 64}_ckpt_{I}_of_10
rlvr_llama1_warmstart_bleu_alma_rbz_{128, 256}_ckpt_{i}_of_10
-
rosieyzh/rlvr_llama1_warmstart_bleu_alma_rbz_256_ckpt_1_of_10
1B • Updated -
rosieyzh/rlvr_llama1_warmstart_bleu_alma_rbz_256_ckpt_2_of_10
1B • Updated • 4 -
rosieyzh/rlvr_llama1_warmstart_bleu_alma_rbz_256_ckpt_3_of_10
1B • Updated -
rosieyzh/rlvr_llama1_warmstart_bleu_alma_rbz_256_ckpt_4_of_10
1B • Updated
rosieyzh/sft_llama1_alma_lr_1e-5_cosine_bsz_{64,128}_ckpt_{i}_of_5
-
rosieyzh/sft_llama1_alma_lr_1e-5_cosine_bsz_64_ckpt_1_of_5
1B • Updated • 4 -
rosieyzh/sft_llama1_alma_lr_1e-5_cosine_bsz_64_ckpt_2_of_5
1B • Updated • 3 -
rosieyzh/sft_llama1_alma_lr_1e-5_cosine_bsz_64_ckpt_3_of_5
1B • Updated • 4 -
rosieyzh/sft_llama1_alma_lr_1e-5_cosine_bsz_64_ckpt_4_of_5
1B • Updated • 4
rosieyzh/rlvr_qwen15_code200_rbz_{32, 64}_ckpt_{I}_of_10
Pretrained models from scratch used in "Echo Chamber: RL Post-training Amplifies Behaviors Learned in Pretraining".
OLMo 1B model pretrained with Algebraic Stack, FineMath3, TinyGSM, and OpenMathInstruct2. Includes checkpoints from doing PPO using GSM8K train.
Datasets used in "Understanding the Design Space and Cross-Modality Transfer for Vision-Language Models"
rosieyzh/sft_qwen15_swallow_lr_1e-5_cosine_bsz_{64,128}_ckpt_{i}_of_5
-
rosieyzh/sft_qwen15_swallow_lr_1e-5_cosine_bsz_128_ckpt_1_of_5
2B • Updated • 9 -
rosieyzh/sft_qwen15_swallow_lr_1e-5_cosine_bsz_128_ckpt_2_of_5
2B • Updated • 8 -
rosieyzh/sft_qwen15_swallow_lr_1e-5_cosine_bsz_128_ckpt_3_of_5
2B • Updated • 6 -
rosieyzh/sft_qwen15_swallow_lr_1e-5_cosine_bsz_128_ckpt_4_of_5
2B • Updated • 6
rosieyzh/rlvr_llama1_warmstart_rouge_xsum_rbz_{32,64}_ckpt_{i}_of_10
rosieyzh/rlvr_llama1_rouge_xsum_rbz_{32,64}_ckpt_{i}_of_10
rosieyzh/sft_llama1_xsum_lr_1e-5_cosine_bsz_{64,128}_ckpt_{i}_of_5
rosieyzh/rlvr_qwen15_warmstart_gsm8k_rbz_{32, 64}_ckpt_{I}_of_10
rosieyzh/rlvr_qwen15_gsm8k_rbz_{32, 64}_ckpt_{I}_of_10
rlvr_llama1_warmstart_bleu_alma_rbz_{128, 256}_ckpt_{i}_of_10
-
rosieyzh/rlvr_llama1_warmstart_bleu_alma_rbz_256_ckpt_1_of_10
1B • Updated -
rosieyzh/rlvr_llama1_warmstart_bleu_alma_rbz_256_ckpt_2_of_10
1B • Updated • 4 -
rosieyzh/rlvr_llama1_warmstart_bleu_alma_rbz_256_ckpt_3_of_10
1B • Updated -
rosieyzh/rlvr_llama1_warmstart_bleu_alma_rbz_256_ckpt_4_of_10
1B • Updated
rlvr_llama1_bleu_alma_rbz_{128,256}_ckpt_{i}_of_10
128: [7, 12, 21, 36, 62, 106, 182, 313, 535, 917]
256: [3, 6, 10, 18, 31, 53, 91, 156, 267, 458]
rosieyzh/sft_llama1_alma_lr_1e-5_cosine_bsz_{64,128}_ckpt_{i}_of_5
-
rosieyzh/sft_llama1_alma_lr_1e-5_cosine_bsz_64_ckpt_1_of_5
1B • Updated • 4 -
rosieyzh/sft_llama1_alma_lr_1e-5_cosine_bsz_64_ckpt_2_of_5
1B • Updated • 3 -
rosieyzh/sft_llama1_alma_lr_1e-5_cosine_bsz_64_ckpt_3_of_5
1B • Updated • 4 -
rosieyzh/sft_llama1_alma_lr_1e-5_cosine_bsz_64_ckpt_4_of_5
1B • Updated • 4
rosieyzh/rlvr_qwen15_warmstart_code200_rbz_{32, 64}_ckpt_{I}_of_10
-
rosieyzh/rlvr_qwen15_warmstart_code200_rbz_32_ckpt_1_of_10
2B • Updated • 3 -
rosieyzh/rlvr_qwen15_warmstart_code200_rbz_32_ckpt_2_of_10
2B • Updated • 4 -
rosieyzh/rlvr_qwen15_warmstart_code200_rbz_32_ckpt_3_of_10
2B • Updated • 3 -
rosieyzh/rlvr_qwen15_warmstart_code200_rbz_32_ckpt_4_of_10
2B • Updated • 3
rosieyzh/rlvr_qwen15_code200_rbz_{32, 64}_ckpt_{I}_of_10
rosieyzh/sft_qwen15_code200_lr_{1e-5, 5e-6}_{cosine, constant}_bsz_{64, 128}_ckpt_{i}_of_5
-
rosieyzh/sft_qwen15_code200_lr_1e-5_cosine_bsz_64_ckpt_1_of_5
2B • Updated • 7 -
rosieyzh/sft_qwen15_code200_lr_1e-5_cosine_bsz_64_ckpt_2_of_5
2B • Updated • 8 -
rosieyzh/sft_qwen15_code200_lr_1e-5_cosine_bsz_64_ckpt_3_of_5
2B • Updated • 8 -
rosieyzh/sft_qwen15_code200_lr_1e-5_cosine_bsz_64_ckpt_4_of_5
2B • Updated • 7
Pretrained models from scratch used in "Echo Chamber: RL Post-training Amplifies Behaviors Learned in Pretraining".
OLMo 1B model pretrained with Algebraic Stack, FineMath3, TinyGSM, OMI1, and OMI2. Includes checkpoints from doing PPO using GSM8K train.
-
rosieyzh/OLMo-1B-as_fm3_tg_omi1_omi2_ppo
Text Generation • 1B • Updated • 2 -
rosieyzh/OLMo-1B-as_fm3_tg_omi1_omi2_episode1
Text Generation • 1B • Updated • 2 -
rosieyzh/OLMo-1B-as_fm3_tg_omi1_omi2_episode2
Text Generation • 1B • Updated • 3 -
rosieyzh/OLMo-1B-as_fm3_tg_omi1_omi2_episode3
Text Generation • 1B • Updated • 2
OLMo 1B model pretrained with Algebraic Stack, FineMath3, TinyGSM, and OpenMathInstruct2. Includes checkpoints from doing PPO using GSM8K train.