update

Dockerfile

@@ -1,3 +1,4 @@
+# syntax=docker/dockerfile:1.4
 FROM docker.io/nvidia/cuda:11.7.1-cudnn8-devel-ubuntu20.04
 ENV TZ=Asia/Tokyo
 RUN ln -snf /usr/share/zoneinfo/$TZ /etc/localtime && echo $TZ > /etc/timezone
@@ -9,7 +10,8 @@ RUN sed -i 's http://deb.debian.org http://cdn-aws.deb.debian.org g' /etc/apt/so
         make build-essential libssl-dev zlib1g-dev \
         libbz2-dev libreadline-dev libsqlite3-dev wget curl llvm \
         libncursesw5-dev xz-utils tk-dev libxml2-dev libxmlsec1-dev libffi-dev liblzma-dev git-lfs \
-        ffmpeg libsm6 libxext6 cmake libgl1-mesa-glx && \
+        ffmpeg libsm6 libxext6 cmake libgl1-mesa-glx \
+        python3.9-dev && \
     rm -rf /var/lib/apt/lists/* && \
     git lfs install
 RUN curl https://pyenv.run | bash

app.py

@@ -2,6 +2,7 @@ from typing import Optional
 
 import gradio as gr
 import torch
+import transformers
 from peft import PeftModel
 from transformers import GenerationConfig
 
@@ -10,7 +11,8 @@ print("starting server ...")
 assert (
     "LlamaTokenizer" in transformers._import_structure["models.llama"]
 ), "LLaMA is now in HuggingFace's main branch.\nPlease reinstall it: pip uninstall transformers && pip install git+https://github.com/huggingface/transformers.git"
-from transformers import LlamaTokenizer, LlamaForCausalLM
+from transformers import LlamaForCausalLM
+from transformers import LlamaTokenizer
 
 BASE_MODEL = "decapoda-research/llama-13b-hf"
 LORA_WEIGHTS = "izumi-lab/llama-13b-japanese-lora-v0-1ep"
@@ -35,9 +37,7 @@ if device == "cuda":
         torch_dtype=torch.float16,
         device_map="auto",
     )
-    model = PeftModel.from_pretrained(
-        model, LORA_WEIGHTS, torch_dtype=torch.float16
-    )
+    model = PeftModel.from_pretrained(model, LORA_WEIGHTS, torch_dtype=torch.float16)
 elif device == "mps":
     model = LlamaForCausalLM.from_pretrained(
         BASE_MODEL,

model_pull.py

@@ -1,10 +1,12 @@
 import torch
+import transformers
 from peft import PeftModel
 
 assert (
     "LlamaTokenizer" in transformers._import_structure["models.llama"]
 ), "LLaMA is now in HuggingFace's main branch.\nPlease reinstall it: pip uninstall transformers && pip install git+https://github.com/huggingface/transformers.git"
-from transformers import LlamaTokenizer, LlamaForCausalLM
+from transformers import LlamaForCausalLM
+from transformers import LlamaTokenizer
 
 BASE_MODEL = "decapoda-research/llama-13b-hf"
 LORA_WEIGHTS = "izumi-lab/llama-13b-japanese-lora-v0-1ep"