Datasets and models associated with the paper "Large-Scale Data Selection for Instruction Tuning" (https://arxiv.org/abs/2503.01807)
Hamish Ivison
hamishivi
AI & ML interests
NLP :)
Recent Activity
updated
a dataset
6 days ago
hamishivi/tulu_3_rewritten_400k_string_f1_only_v2_nocode_all_filtered_qwen2_5_openthoughts2
updated
a dataset
6 days ago
hamishivi/virtuoussy_multi_subject_rlvr
updated
a dataset
7 days ago
hamishivi/llama_nemotron_post_training_sft_science
Organizations
Llama 3 models trained on the tulu dataset, following https://arxiv.org/abs/2311.10702 (tulu 2) and https://arxiv.org/abs/2406.09279 (tulu 2.5).
The set of models associated with the Tulu V2 technical report.
Models associated with the paper "TESS-2: A Large-Scale, Generalist Diffusion Language Model". Code: https://github.com/hamishivi/tess-2
a small run at 7b scale with ppo, following the unpacking dpo and ppo paper.
The set of models associated with the paper "How Far Can Camels Go? Exploring the State of Instruction Tuning on Open Resources".
Datasets and models associated with the paper "Large-Scale Data Selection for Instruction Tuning" (https://arxiv.org/abs/2503.01807)
Models associated with the paper "TESS-2: A Large-Scale, Generalist Diffusion Language Model". Code: https://github.com/hamishivi/tess-2
Llama 3 models trained on the tulu dataset, following https://arxiv.org/abs/2311.10702 (tulu 2) and https://arxiv.org/abs/2406.09279 (tulu 2.5).
a small run at 7b scale with ppo, following the unpacking dpo and ppo paper.
The set of models associated with the Tulu V2 technical report.
The set of models associated with the paper "How Far Can Camels Go? Exploring the State of Instruction Tuning on Open Resources".