update examples

examples/README.md

@@ -23,8 +23,8 @@ examples/
 │   ├── single_node.sh
 │   └── multi_node.sh
 ├── merge_lora/
-│   ├── merge.sh
-│   └── quantize.sh
+│   ├── merge.sh: Merge LoRA weights
+│   └── quantize.sh: Quantize with AutoGPTQ
 ├── inference/
 │   ├── cli_demo.sh
 │   ├── api_demo.sh

examples/extras/fsdp_qlora/README.md

@@ -1,5 +0,0 @@
-```bash
-pip install "transformers>=4.39.1"
-pip install "accelerate>=0.28.0"
-pip install "bitsandbytes>=0.43.0"
-```

examples/extras/fsdp_qlora/sft.sh

@@ -1,5 +1,9 @@
 #!/bin/bash
 
+pip install "transformers>=4.39.1"
+pip install "accelerate>=0.28.0"
+pip install "bitsandbytes>=0.43.0"
+
 CUDA_VISIBLE_DEVICES=0,1 accelerate launch \
     --config_file ../accelerate/fsdp_config.yaml \
     ../../src/train_bash.py \

examples/lora_single_gpu/README.md

@@ -1,9 +0,0 @@
-Usage:
-
-- `pretrain.sh`: do pre-train (optional)
-- `sft.sh`: do supervised fine-tuning
-- `reward.sh`: do reward modeling (must after sft.sh)
-- `ppo.sh`: do PPO training (must after sft.sh and reward.sh)
-- `dpo.sh`: do DPO training (must after sft.sh)
-- `orpo.sh`: do ORPO training
-- `predict.sh`: do predict (must after sft.sh and dpo.sh)

examples/merge_lora/README.md

@@ -1,13 +0,0 @@
-> [!WARNING]
-> Merging LoRA weights into a quantized model is not supported.
-
-> [!TIP]
-> Use `--model_name_or_path path_to_model` solely to use the exported model or model fine-tuned in full/freeze mode.
->
-> Use `CUDA_VISIBLE_DEVICES=0`, `--export_quantization_bit 4` and `--export_quantization_dataset data/c4_demo.json` to quantize the model with AutoGPTQ after merging the LoRA weights.
-
-
-Usage:
-
-- `merge.sh`: merge the lora weights
-- `quantize.sh`: quantize the model with AutoGPTQ (must after merge.sh, optional)