Added Dockerfile and docker-compose.yml (#207)

* Added Dockerfile for inference

* Added instructions for Dockerfile

* Update README.md

* Update README.md

* Update README.md

* Pass env through Dockerfile

* Added docker compose setup and instructions

* Added more environment options

* Set a safer default mount point

* add docker-compose changes

* Added Dockerfile for inference

* Added instructions for Dockerfile

* Update README.md

* Update README.md

* Update README.md

* Pass env through Dockerfile

* Added docker compose setup and instructions

* Added more environment options

* Set a safer default mount point

* add to gitignore, update to new generate.py

* add docker ignore, simplify docker compose file

* add back missing requirements

* Adjustments to compose and generate.py, added Docker to README.md

* Linting adjust to Black

* Adjusting import linting

* Update README.md

* Update README.md

* Removed comment by original Dockerfile creator.

Comment not necessary.

* cleanup README

Co-authored-by: Francesco Saverio Zuppichini <zuppif@usi.ch>

---------

Co-authored-by: Francesco Saverio Zuppichini <zuppif@usi.ch>
Co-authored-by: Chris Alexiuk <c.s.alexiuk@gmail.com>
Co-authored-by: ElRoberto538 <>
Co-authored-by: Sam Sipe <samsipe@gmail.com>
Co-authored-by: Eric J. Wang <eric.james.wang@gmail.com>

.dockerignore

@@ -0,0 +1,4 @@
+.venv
+.github
+.vscode
+.docker-compose.yml

.gitignore

@@ -11,4 +11,5 @@ wandb
 evaluate.py
 test_data.json
 todo.txt
-.vscode/
+.venv
+.vscode

Dockerfile

@@ -0,0 +1,18 @@
+FROM nvidia/cuda:11.8.0-devel-ubuntu22.04
+
+ARG DEBIAN_FRONTEND=noninteractive
+
+RUN apt-get update && apt-get install -y \
+    git \
+    curl \
+    software-properties-common \
+    && add-apt-repository ppa:deadsnakes/ppa \
+    && apt install -y python3.10 \
+    && rm -rf /var/lib/apt/lists/*
+WORKDIR /workspace
+COPY requirements.txt requirements.txt
+RUN curl -sS https://bootstrap.pypa.io/get-pip.py | python3.10 \
+    && python3.10 -m pip install -r requirements.txt \
+    && python3.10 -m pip install numpy --pre torch --force-reinstall --index-url https://download.pytorch.org/whl/nightly/cu118
+COPY . .
+ENTRYPOINT [ "python3.10"]

README.md

@@ -15,25 +15,13 @@ as well as Tim Dettmers' [bitsandbytes](https://github.com/TimDettmers/bitsandby
 
 Without hyperparameter tuning, the LoRA model produces outputs comparable to the Stanford Alpaca model. (Please see the outputs included below.) Further tuning might be able to achieve better performance; I invite interested users to give it a try and report their results.
 
-## Setup
+### Local Setup
 
 1. Install dependencies
 
-    ```bash
-    pip install -r requirements.txt
-    ```
-
-1. Set environment variables, or modify the files referencing `BASE_MODEL`:
-
-    ```bash
-    # Files referencing `BASE_MODEL`
-    # export_hf_checkpoint.py
-    # export_state_dict_checkpoint.py
-
-    export BASE_MODEL=decapoda-research/llama-7b-hf
-    ```
-
-    Both `finetune.py` and `generate.py` use `--base_model` flag as shown further below.
+   ```bash
+   pip install -r requirements.txt
+   ```
 
 1. If bitsandbytes doesn't work, [install it from source.](https://github.com/TimDettmers/bitsandbytes/blob/main/compile_from_source.md) Windows users can follow [these instructions](https://github.com/tloen/alpaca-lora/issues/17).
 
@@ -94,6 +82,49 @@ They should help users
 who want to run inference in projects like [llama.cpp](https://github.com/ggerganov/llama.cpp)
 or [alpaca.cpp](https://github.com/antimatter15/alpaca.cpp).
 
+### Docker Setup & Inference
+
+1. Build the container image:
+
+```bash
+docker build -t alpaca-lora .
+```
+
+2. Run the container (you can also use `finetune.py` and all of its parameters as shown above for training):
+
+```bash
+docker run --gpus=all --shm-size 64g -p 7860:7860 -v ${HOME}/.cache:/root/.cache --rm alpaca-lora generate.py \
+    --load_8bit \
+    --base_model 'decapoda-research/llama-7b-hf' \
+    --lora_weights 'tloen/alpaca-lora-7b'
+```
+
+3. Open `https://localhost:7860` in the browser
+
+### Docker Compose Setup & Inference
+
+1. (optional) Change desired model and weights under `environment` in the `docker-compose.yml`
+
+2. Build and run the container
+
+```bash
+docker-compose up -d --build
+```
+
+3. Open `https://localhost:7860` in the browser
+
+4. See logs:
+
+```bash
+docker-compose logs -f
+```
+
+5. Clean everything up:
+
+```bash
+docker-compose down --volumes --rmi all
+```
+
 ### Notes
 
 - We can likely improve our model performance significantly if we had a better dataset. Consider supporting the [LAION Open Assistant](https://open-assistant.io/) effort to produce a high-quality dataset for supervised fine-tuning (or bugging them to release their data).
@@ -110,9 +141,7 @@ or [alpaca.cpp](https://github.com/antimatter15/alpaca.cpp).
   - 7B:
     - <https://huggingface.co/tloen/alpaca-lora-7b>
     - <https://huggingface.co/samwit/alpaca7B-lora>
-    - 🤖 <https://huggingface.co/nomic-ai/gpt4all-lora>
     - 🇧🇷 <https://huggingface.co/22h/cabrita-lora-v0-1>
-    - 🇨🇳 <https://huggingface.co/ziqingyang/chinese-alpaca-lora-7b>
     - 🇨🇳 <https://huggingface.co/qychen/luotuo-lora-7b-0.1>
     - 🇯🇵 <https://huggingface.co/kunishou/Japanese-Alapaca-LoRA-7b-v0>
     - 🇫🇷 <https://huggingface.co/bofenghuang/vigogne-lora-7b>
@@ -131,9 +160,6 @@ or [alpaca.cpp](https://github.com/antimatter15/alpaca.cpp).
     - <https://huggingface.co/baseten/alpaca-30b>
     - <https://huggingface.co/chansung/alpaca-lora-30b>
     - 🇯🇵 <https://huggingface.co/kunishou/Japanese-Alapaca-LoRA-30b-v0>
-    - 🇰🇷 <https://huggingface.co/beomi/KoAlpaca-30B-LoRA>
-  - 65B:
-    - 🇰🇷 <https://huggingface.co/beomi/KoAlpaca-65B-LoRA>
 - [alpaca-native](https://huggingface.co/chavinlo/alpaca-native), a replication using the original Alpaca code
 
 ### Example outputs

docker-compose.yml

@@ -0,0 +1,28 @@
+version: '3'
+
+services:
+  alpaca-lora:
+    build:
+      context: ./
+      dockerfile: Dockerfile
+      args:
+        BUILDKIT_INLINE_CACHE: "0"
+    image: alpaca-lora
+    shm_size: '64gb'
+    command: generate.py --load_8bit --base_model $BASE_MODEL --lora_weights 'tloen/alpaca-lora-7b'
+    restart: unless-stopped
+    volumes:
+      - alpaca-lora:/root/.cache # Location downloaded weights will be stored
+    ports:
+      - 7860:7860
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [ gpu ]
+
+volumes:
+  alpaca-lora:
+    name: alpaca-lora

generate.py

@@ -1,3 +1,4 @@
+import os
 import sys
 
 import fire
@@ -29,6 +30,7 @@ def main(
     server_name: str = "127.0.0.1",  # Allows to listen on all interfaces by providing '0.0.0.0'
     share_gradio: bool = False,
 ):
+    base_model = base_model or os.environ.get("BASE_MODEL", "")
     assert (
         base_model
     ), "Please specify a --base_model, e.g. --base_model='decapoda-research/llama-7b-hf'"
@@ -146,7 +148,7 @@ def main(
         ],
         title="🦙🌲 Alpaca-LoRA",
         description="Alpaca-LoRA is a 7B-parameter LLaMA model finetuned to follow instructions. It is trained on the [Stanford Alpaca](https://github.com/tatsu-lab/stanford_alpaca) dataset and makes use of the Huggingface LLaMA implementation. For more information, please visit [the project's website](https://github.com/tloen/alpaca-lora).",  # noqa: E501
-    ).launch(server_name=server_name, share=share_gradio)
+    ).launch(server_name="0.0.0.0", share=share_gradio)
     # Old testing code follows.
 
     """

requirements.txt

@@ -1,5 +1,6 @@
 accelerate
 appdirs
+loralib
 bitsandbytes
 black
 black[jupyter]
@@ -7,5 +8,6 @@ datasets
 fire
 git+https://github.com/huggingface/peft.git
 git+https://github.com/huggingface/transformers.git
+sentencepiece
 gradio
 sentencepiece