diff --git a/scripts/performance/configs/deepseek/deepseek_v3_llm_pretrain.yaml b/scripts/performance/configs/deepseek/deepseek_v3_llm_pretrain.yaml index 7f56df49ac..044d99536e 100644 --- a/scripts/performance/configs/deepseek/deepseek_v3_llm_pretrain.yaml +++ b/scripts/performance/configs/deepseek/deepseek_v3_llm_pretrain.yaml @@ -24,6 +24,9 @@ ConfigContainer: train: train_iters: 100 eval_iters: 0 + + scheduler: + lr_warmup_iters: 10 rerun_state_machine: check_for_nan_in_loss: false diff --git a/scripts/performance/configs/llama3/llama3_70b_llm_pretrain.yaml b/scripts/performance/configs/llama3/llama3_70b_llm_pretrain.yaml index d59a11c3a8..11588a4231 100644 --- a/scripts/performance/configs/llama3/llama3_70b_llm_pretrain.yaml +++ b/scripts/performance/configs/llama3/llama3_70b_llm_pretrain.yaml @@ -25,6 +25,9 @@ ConfigContainer: train_iters: 100 eval_iters: 0 + scheduler: + lr_warmup_iters: 10 + rerun_state_machine: check_for_nan_in_loss: false diff --git a/scripts/performance/configs/llama3/llama3_8b_llm_pretrain.yaml b/scripts/performance/configs/llama3/llama3_8b_llm_pretrain.yaml index c987e0b712..56f21c5698 100644 --- a/scripts/performance/configs/llama3/llama3_8b_llm_pretrain.yaml +++ b/scripts/performance/configs/llama3/llama3_8b_llm_pretrain.yaml @@ -20,6 +20,9 @@ ConfigContainer: train: train_iters: 100 eval_iters: 0 + + scheduler: + lr_warmup_iters: 10 rerun_state_machine: check_for_nan_in_loss: false diff --git a/scripts/performance/configs/llama31/llama31_405b_llm_pretrain.yaml b/scripts/performance/configs/llama31/llama31_405b_llm_pretrain.yaml index d904c73e28..4a6211d8b4 100644 --- a/scripts/performance/configs/llama31/llama31_405b_llm_pretrain.yaml +++ b/scripts/performance/configs/llama31/llama31_405b_llm_pretrain.yaml @@ -25,6 +25,9 @@ ConfigContainer: train: train_iters: 100 eval_iters: 0 + + scheduler: + lr_warmup_iters: 10 rerun_state_machine: check_for_nan_in_loss: false diff --git a/scripts/performance/configs/qwen3/qwen3_235b_a22b_llm_pretrain.yaml b/scripts/performance/configs/qwen3/qwen3_235b_a22b_llm_pretrain.yaml index 20657c4f5a..9463caccb0 100644 --- a/scripts/performance/configs/qwen3/qwen3_235b_a22b_llm_pretrain.yaml +++ b/scripts/performance/configs/qwen3/qwen3_235b_a22b_llm_pretrain.yaml @@ -25,6 +25,9 @@ ConfigContainer: train_iters: 100 eval_iters: 0 + scheduler: + lr_warmup_iters: 10 + rerun_state_machine: check_for_nan_in_loss: false diff --git a/scripts/performance/configs/qwen3/qwen3_30b_a3b_llm_pretrain.yaml b/scripts/performance/configs/qwen3/qwen3_30b_a3b_llm_pretrain.yaml index 92d5d529ee..55ec607e43 100644 --- a/scripts/performance/configs/qwen3/qwen3_30b_a3b_llm_pretrain.yaml +++ b/scripts/performance/configs/qwen3/qwen3_30b_a3b_llm_pretrain.yaml @@ -24,6 +24,9 @@ ConfigContainer: train: train_iters: 100 eval_iters: 0 + + scheduler: + lr_warmup_iters: 10 rerun_state_machine: check_for_nan_in_loss: false