gpjt/8xa100m40-rmsnorm-1-hand-layernorm
Text Generation • 0.2B • Updated • 33
🔄 In a Training Loop
Giles Thomas
gpjt
AI & ML interests
Doing my best to speedrun 20 years of AI research. YMMV
Recent Activity
updated a model 7 days ago
gpjt/8xa100m40-lore-11-output-only-corrected-smart-init updated a model 7 days ago
gpjt/8xa100m40-lore-10-both-corrected-smart-init updated a model 7 days ago
gpjt/8xa100m40-lore-9-new-seed-output-onlyOrganizations
None yet
gpjt/8xa100m40-rmsnorm-4-pytorch-rmsnorm
Text Generation • 0.2B • Updated • 36
gpjt/8xa100m40-rmsnorm-3-hand-rmsnorm
Text Generation • 0.2B • Updated • 28
gpjt/8xa100m40-rmsnorm-2-pytorch-layernorm
Text Generation • 0.2B • Updated • 30
gpjt/8xa100m40-lore-7-new-seed-input-only
Text Generation • 0.1B • Updated • 233
gpjt/8xa100m40-lore-6-new-seed-baseline-disabled
Text Generation • 0.2B • Updated • 226
gpjt/8xa100m40-lore-5-output-only
Text Generation • 0.1B • Updated • 238
gpjt/8xa100m40-lore-4-input-only
Text Generation • 0.1B • Updated • 242
gpjt/8xa100m40-lore-3-both-smart-init
Text Generation • 0.1B • Updated • 242
gpjt/8xa100m40-lore-2-both-ordinary-init
Text Generation • 0.1B • Updated • 225
gpjt/8xa100m40-lore-1-baseline-disabled
Text Generation • 0.2B • Updated • 199
gpjt/8xa100m40-lore-11-output-only-corrected-smart-init
Text Generation • 0.1B • Updated • 211
gpjt/8xa100m40-lore-10-both-corrected-smart-init
Text Generation • 0.1B • Updated • 190
gpjt/8xa100m40-lore-9-new-seed-output-only
Text Generation • 0.1B • Updated • 264
gpjt/8xa100m40-lore-8-new-seed-both-smart-init
Text Generation • 0.1B • Updated • 242
gpjt/jax-with-mha-bias-fw-fwedu-simplewiki-DEPRECATED
Text Generation • 0.2B • Updated • 134
gpjt/jax-with-mha-bias-fw-fwedu-5050-DEPRECATED
Text Generation • 0.2B • Updated • 126 • 1
gpjt/jax-with-mha-bias-fw-fwedu-simplewiki
Text Generation • 0.2B • Updated • 142
gpjt/jax-with-mha-bias-fw-fwedu-5050
Text Generation • 0.2B • Updated • 114
gpjt/jax-with-mha-bias-openwebtext
Text Generation • 0.2B • Updated • 74
gpjt/1xrtx3090-moe-1
Text Generation • 0.5B • Updated • 80
gpjt/jax-with-mha-bias-fineweb-edu
Text Generation • 0.2B • Updated • 68
gpjt/jax-with-mha-bias-larger-chinchilla-1
Text Generation • 0.3B • Updated • 36
gpjt/jax-with-mha-bias-larger-chinchilla-2
Text Generation • 0.2B • Updated • 29
gpjt/jax-with-mha-bias-no-dropout-2-epoch
Text Generation • 0.2B • Updated • 32
gpjt/jax-with-mha-bias-no-dropout-extended
Text Generation • 0.2B • Updated • 36
gpjt/jax-no-mha-bias-no-dropout
Text Generation • 0.2B • Updated • 35
gpjt/jax-with-mha-bias-no-dropout
Text Generation • 0.2B • Updated • 28
gpjt/jax-no-mha-bias-with-dropout
Text Generation • 0.2B • Updated • 24
gpjt/1xrtx3090-stacked-interventions
Text Generation • 0.2B • Updated • 9