Update README.md

README.md

@@ -188,13 +188,17 @@ If you *really* want to use `<|im_start|>` and `<|im_end|>`, just update your `t
 
 An example for mistral-7b:
 
+*Note: I actually used my fork of [qlora](https://github.com/jondurbin/qlora)'s `train.py` for this, but I'm porting it to a minified version here, not tested yet!*
+
+*More notes: I stopped the SFT phase around 50% because of budget constraints.*
+
 ```bash
 export BASE_DIR=/workspace
 export WANDB_API_KEY=[redacted]
 export WANDB_PROJECT=bagel-7b-v0.1
 
 # Run the pretraining.
-accelerate launch -m bagel.tune.sft \
+accelerate launch bagel/tune/sft.py \
   --model_name_or_path $BASE_DIR/mistral-7b \
   --final_output_dir $BASE_DIR/$WANDB_PROJECT \
   --output_dir $BASE_DIR/$WANDB_PROJECT-workdir \
@@ -266,7 +270,7 @@ export BASE_DIR=/mnt/data
 export WANDB_API_KEY=[redacted]
 export WANDB_PROJECT=bagel-dpo-7b-v0.1
 
-accelerate launch -m bagel.tune.dpo \
+accelerate launch bagel/tune/dpo.py \
   --model_name_or_path bagel-7b-v0.1 \
   --learning_rate 3e-7 \
   --per_device_train_batch_size 2 \