From ebcdd9d255920ba23cb393a7aa4497a2b886f259 Mon Sep 17 00:00:00 2001 From: Sahil Jain Date: Mon, 31 Mar 2025 11:34:10 -0700 Subject: [PATCH] Grammar nit Signed-off-by: Sahil Jain --- README.md | 2 +- examples/configs/sft.yaml | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index aeab19f33c..4b0b5cbba9 100644 --- a/README.md +++ b/README.md @@ -66,7 +66,7 @@ The default SFT experiment is configured to run on a single GPU. To launch the e uv run python examples/run_sft.py ``` -This trains `Llama3.2-1B` on one GPU using SQUAD dataset. +This trains `Llama3.2-1B` on one GPU using the SQUAD dataset. If you have access to more GPUs, you can update the experiment accordingly. To run on 8 GPUs, we update the cluster configuration. We also switch to an 8B Llama base model and increase the batch size: diff --git a/examples/configs/sft.yaml b/examples/configs/sft.yaml index 793ebb6285..e4b116a351 100644 --- a/examples/configs/sft.yaml +++ b/examples/configs/sft.yaml @@ -37,7 +37,7 @@ data: logger: log_dir: "logs" # Base directory for all logs - wandb_enabled: true # Make sure you do ``wandb login [Your API key]'' before run + wandb_enabled: true # Make sure you do a ``wandb login [Your API key]'' before running tensorboard_enabled: true monitor_gpus: false # If true, will monitor GPU usage and log to wandb and/or tensorboard wandb: