21 days ago
-
add LLaMA-65B (the run that reproduces the paper, e2e lr 1e-6)
23 days ago
-
LLaMA-2-7B-chat: keep only the released (lr 2e-5) run
-
drop the -ft2e-6 alias and Llama-3.1
-
DeepSeek: keep the lr 2e-6 run as the released model
-
Upload dbw-qwen2.5-7b-w1.61g128.pt with huggingface_hub
-
Upload dbw-mistral-7b-w1.61g128.pt with huggingface_hub
-
Upload dbw-gemma-7b-w1.61g128.pt with huggingface_hub
-
Upload dbw-Llama-7b-w1.61g128.pt with huggingface_hub
-
Upload dbw-Llama-30b-w1.61g128.pt with huggingface_hub
-
Upload dbw-Llama-3.1-8b-w1.61g128.pt with huggingface_hub
-
Upload dbw-Llama-3-8b-w1.61g128.pt with huggingface_hub
-
Upload dbw-Llama-2-7b-w1g128.pt with huggingface_hub
-
Upload dbw-Llama-2-7b-w1.61g128.pt with huggingface_hub
-
Upload dbw-Llama-2-7b-chat-w1.61g128.pt with huggingface_hub
-
Upload dbw-Llama-2-7b-chat-w1.61g128-ft2e-6.pt with huggingface_hub
-
Upload dbw-Llama-2-70b-w1.61g128.pt with huggingface_hub
-
Upload dbw-Llama-2-13b-w1.61g128.pt with huggingface_hub
-
Upload dbw-Llama-13b-w1.61g128.pt with huggingface_hub
-
Upload dbw-DeepSeek-R1-Distill-Llama-8B-w1.61g128.pt with huggingface_hub
-
Upload dbw-DeepSeek-R1-Distill-Llama-8B-w1.61g128-ft2e-6.pt with huggingface_hub
-
Upload dbw-Llama-2-7b-w0.5g128.pt with huggingface_hub
-
initial commit
