fix3rdparty--rebase (#239)

Author: CeciliaWwq

docker/Dockerfile

@@ -29,7 +29,10 @@ FROM itrex-base as devel
 ENV DEBIAN_FRONTEND=noninteractive
 
 ENV LANG C.UTF-8
-ARG PYTHON=python3.10
+ARG PYTHON=python3.10
+
+RUN mkdir -p /app/intel-extension-for-transformers
+WORKDIR /app/intel-extension-for-transformers
 
 RUN apt-get update && apt-get install -y --no-install-recommends --fix-missing \
  autoconf \
@@ -42,9 +45,7 @@ RUN apt-get update && apt-get install -y --no-install-recommends --fix-missing \
  apt-get autoremove -y && \
  rm -rf /var/lib/apt/lists/*
 
-COPY . /tmp/intel-extension-for-transformers
-
-WORKDIR /tmp/intel-extension-for-transformers
+COPY . /app/intel-extension-for-transformers
 
 RUN ${PYTHON} -m pip install -r requirements.txt --no-cache-dir -f https://download.pytorch.org/whl/cpu/torch_stable.html
 

intel_extension_for_transformers/llm/runtime/deprecated/executor/src/i_malloc.cpp

@@ -99,6 +99,7 @@ static void mempool_enlarge(struct malloc_mempool* pool, size_t increase_size) {
  if (pool->start_addr != NULL) {
 #ifdef _WIN32
  free(pool->start_addr);
+ pool->start_addr = NULL;
  int ret = 0;
 #else
  int ret = munmap(pool->start_addr, pool->total_size);

workflows/chatbot/demo/basic_frontend/fastchat/eval/webpage/index.html

@@ -138,14 +138,13 @@ <h2 class="text-center mb-5">Who's GPT-4's favorite? Battles between State-of-th
  </div>
  </div>
 
- <!-- Marked.js -->
- <script src="https://cdn.jsdelivr.net/npm/marked@4.3.0/lib/marked.umd.min.js"></script>
- <!-- Bootstrap and Popper.js JavaScript dependencies -->
- <script src="https://code.jquery.com/jquery-3.5.1.slim.min.js"></script>
- <script src="https://cdn.jsdelivr.net/npm/@popperjs/core@2.11.6/dist/umd/popper.min.js"></script>
- <script src="https://maxcdn.bootstrapcdn.com/bootstrap/4.5.2/js/bootstrap.min.js"></script>
-
- <script src="script.js"></script>
+ <!-- Marked.js -->
+ <script src="https://cdn.jsdelivr.net/npm/marked@4.3.0/lib/marked.umd.min.js" integrity="sha384-b7a7929253062ca68a1b4f243308ad4200e2594c9e27453031250c4ffc6b34360f7f3bdbb1413c78c712092168bafcd6" crossorigin="anonymous"></script>
+ <!-- Bootstrap and Popper.js JavaScript dependencies -->
+ <script src="https://code.jquery.com/jquery-3.5.1.slim.min.js" integrity="sha384-0df5ddcf686d3c7d25b124ace67093a6e8ffcf2e02f8e1a96a6a05572dfc31506713e21b6d56147b0f8eac25da4647e3" crossorigin="anonymous"></script>
+ <script src="https://cdn.jsdelivr.net/npm/@popperjs/core@2.11.6/dist/umd/popper.min.js" integrity="sha384-a01a83566333f404cac487a9f6d882c52fd9f5f35f1178830184ee8cc01e040b2316e0994a629b4925274259a1fe3a77" crossorigin="anonymous"></script>
+ <script src="https://maxcdn.bootstrapcdn.com/bootstrap/4.5.2/js/bootstrap.min.js" integrity="sha384-07882dd63ac60bb261e008133d2754b4e06f7cef2c86e7f9ec16a086a15f3e5631a1fbcbf3f29411f62224e30c57ead5" crossorigin="anonymous"></script>
+ <script src="script.js"></script>
  <script>
  // Fetch the JSON file
  fetch('data.json')

workflows/chatbot/demo/docker/Dockerfile

@@ -56,7 +56,7 @@ RUN apt-get install default-jdk -y
 
 # Install and upgrade node and npm for frontend
 RUN apt-get install npm -y && \
- npm install -g n && \
+ npm install -g n --ignore-scripts && \
  n stable && \
  hash -r && \
  npm install -g npm@latest

workflows/chatbot/inference/backend/chat/tts.py

@@ -11,7 +11,8 @@
 import intel_extension_for_pytorch as ipex
 import time
 import numpy as np
-from torch.utils.data import DataLoader
+from torch.utils.data import DataLoader
+import tempfile
 
 class TextToSpeech:
  """Convert text to speech with a driven speaker embedding
@@ -28,11 +29,12 @@ def __init__(self):
  self.processor = SpeechT5Processor.from_pretrained("microsoft/speecht5_tts")
  self.device = "cpu"
  self.spk_model_name = "speechbrain/spkrec-xvect-voxceleb"
- self.speaker_model = EncoderClassifier.from_hparams(
- source=self.spk_model_name,
- run_opts={"device": self.device},
- savedir=os.path.join("/tmp", self.spk_model_name)
- )
+ with tempfile.TemporaryFile(dir=os.path.join("/tmp", self.spk_model_name), mode="w+") as file:
+ self.speaker_model = EncoderClassifier.from_hparams(
+ source=self.spk_model_name,
+ run_opts={"device": self.device},
+ savedir=file.name
+ )
  self.vocoder = SpeechT5HifiGan.from_pretrained("microsoft/speecht5_hifigan")
  self.vocoder.eval()
  self.default_speaker_embedding = torch.load('speaker_embeddings/spk_embed_default.pt') # load the default speaker embedding

workflows/chatbot/inference/backend/fastrag/fastrag_service.py

@@ -12,7 +12,8 @@
 import shutil
 import base64
 import os
-import re
+import re
+import tempfile
 from typing import Any, Dict
 from threading import Thread
 from haystack.telemetry import send_event_if_public_demo
@@ -48,19 +49,18 @@
 
 def ask_gm_documents_dense_embedding(folder_path, process_content=False):
  documents = []
- for dirpath, dirnames, filenames in os.walk(folder_path):
- for filename in filenames:
- if filename.endswith(".json"):
- documents = doc_index.d_load_jsonl_file(os.path.join(dirpath, filename), process_content, documents)
- elif filename.endswith(".xlsx"):
- documents = doc_index.d_load_xlsx(os.path.join(dirpath, filename), process_content)
- else:
- print("{} is ignored. Will support this file format soon.".format(filename))
- continue
- doc_index.persist_embedding(documents, "/tmp/ask_gm_dense_retrieval_chinese",
- model_path="shibing624/text2vec-large-chinese")
- doc_index.persist_embedding(documents, "/tmp/ask_gm_dense_retrieval_english",
- model_path="hkunlp/instructor-large")
+ with tempfile.TemporaryDirectory(dir="/tmp/my_subdirectory") as temp_dir:
+ for dirpath, dirnames, filenames in os.walk(folder_path):
+ for filename in filenames:
+ if filename.endswith(".json"):
+ documents = doc_index.d_load_jsonl_file(os.path.join(dirpath, filename), process_content, documents)
+ elif filename.endswith(".xlsx"):
+ documents = doc_index.d_load_xlsx(os.path.join(dirpath, filename), process_content)
+ else:
+ print("{} is ignored. Will support this file format soon.".format(filename))
+ continue
+ doc_index.persist_embedding(documents, temp_dir, model_path="shibing624/text2vec-large-chinese")
+ doc_index.persist_embedding(documents, temp_dir, model_path="hkunlp/instructor-large")
 
 def ask_gm_documents_sparse_embedding(folder_path, process_content=False):
  document_store = ElasticsearchDocumentStore(host="localhost", index="elastic_askgm_sparse",
@@ -141,27 +141,31 @@ def ask_gm_documents_sparse_embedding(folder_path, process_content=False):
 stop_token_ids.append(langchain_tok("。", return_tensors="pt").input_ids)
 stop_token_ids.append(langchain_tok("！", return_tensors="pt").input_ids)
 langchain_tok.pad_token = langchain_tok.eos_token
-langchain_tok.add_special_tokens({'pad_token': '[PAD]'})
-if not os.path.exists("/tmp/young_pat_dense_retrieval"):
- documents = doc_index.d_load_young_pat_xlsx("./doc/young_pat/pat.xlsx", True)
- doc_index.persist_embedding(documents, "/tmp/young_pat_dense_retrieval", model_path="hkunlp/instructor-large")
-
-english_embeddings = HuggingFaceInstructEmbeddings(model_name="hkunlp/instructor-large")
-chinese_embeddings = HuggingFaceInstructEmbeddings(model_name="shibing624/text2vec-base-chinese")
-young_pat_vectordb = Chroma(persist_directory="/tmp/young_pat_dense_retrieval",
- embedding_function=english_embeddings)
-young_pat_dense_retriever = young_pat_vectordb.as_retriever(search_type = "mmr",
- search_kwargs = {"k": 2, "fetch_k": 5})
-
-ask_gm_eng_vectordb = Chroma(persist_directory='/tmp/ask_gm_dense_retrieval_english',
+langchain_tok.add_special_tokens({'pad_token': '[PAD]'})
+with tempfile.TemporaryDirectory(dir="/tmp/my_subdirectory") as temp_dir:
+ if not os.path.exists(temp_dir):
+ documents = doc_index.d_load_young_pat_xlsx("./doc/young_pat/pat.xlsx", True)
+ doc_index.persist_embedding(documents, temp_dir, model_path="hkunlp/instructor-large")
+
+with tempfile.TemporaryDirectory(dir="/tmp/my_subdirectory") as temp_dir:
+ english_embeddings = HuggingFaceInstructEmbeddings(model_name="hkunlp/instructor-large")
+ chinese_embeddings = HuggingFaceInstructEmbeddings(model_name="shibing624/text2vec-base-chinese")
+ 
+ young_pat_vectordb = Chroma(persist_directory=temp_dir,
+ embedding_function=english_embeddings)
+ young_pat_dense_retriever = young_pat_vectordb.as_retriever(search_type="mmr",
+ search_kwargs={"k": 2, "fetch_k": 5})
+
+ ask_gm_eng_vectordb = Chroma(persist_directory=temp_dir,
  embedding_function=english_embeddings)
-ask_gm_eng_retriever = ask_gm_eng_vectordb.as_retriever(search_type = "mmr",
-  search_kwargs = {"k": 2, "fetch_k": 5})
+ ask_gm_eng_retriever = ask_gm_eng_vectordb.as_retriever(search_type="mmr",
+ search_kwargs={"k": 2, "fetch_k": 5})
 
-ask_gm_chn_vectordb = Chroma(persist_directory='/tmp/ask_gm_dense_retrieval_chinese',
+ ask_gm_chn_vectordb = Chroma(persist_directory=temp_dir,
  embedding_function=chinese_embeddings)
-ask_gm_chn_retriever = ask_gm_chn_vectordb.as_retriever(search_type = "mmr",
- search_kwargs = {"k": 2, "fetch_k": 5})
+ ask_gm_chn_retriever = ask_gm_chn_vectordb.as_retriever(search_type="mmr",
+ search_kwargs={"k": 2, "fetch_k": 5})
+
 
 class StopOnTokens(StoppingCriteria):
  def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor, **kwargs) -> bool:
@@ -453,13 +457,15 @@ def query(request: QueryRequest):
  young_pat_pipeline.add_node(component=shaper, name="Shaper", inputs=["Reranker"])
  young_pat_pipeline.add_node(component=prompt, name="Prompter", inputs=["Shaper"])
  result = _process_request(young_pat_pipeline, request)
- elif domain == "Customized":
- if request.blob:
- file_content = base64.b64decode(request.blob)
- random_suffix = str(uuid.uuid4().hex)
- file_path = f"/tmp/customized_doc_{random_suffix}" + request.filename
- with open(file_path, "wb") as f:
- f.write(file_content)
+ elif domain == "Customized":
+if request.blob:
+ file_content = base64.b64decode(request.blob)
+ random_suffix = str(uuid.uuid4().hex)
+ sanitized_filename = os.path.basename(request.filename)
+ file_path = f"/tmp/customized_doc_{random_suffix}_{sanitized_filename}"
+ with open(file_path, "wb") as f:
+ f.write(file_content)
+
 
  if request.filename.endswith("md"):
  converter = MarkdownConverter()