Duplicate from thomas-yanxin/LangChain-ChatLLM

Co-authored-by: thomas Yan <[email protected]>

.gitattributes

@@ -0,0 +1,34 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,14 @@
+---
+title: LangChain ChatLLM
+emoji: ⚡
+colorFrom: green
+colorTo: yellow
+sdk: gradio
+sdk_version: 3.27.0
+app_file: app.py
+pinned: false
+license: apache-2.0
+duplicated_from: thomas-yanxin/LangChain-ChatLLM
+---
+
+Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -0,0 +1,272 @@
+import os
+
+import gradio as gr
+import nltk
+import sentence_transformers
+import torch
+from duckduckgo_search import ddg
+from duckduckgo_search.utils import SESSION
+from langchain.chains import RetrievalQA
+from langchain.document_loaders import UnstructuredFileLoader
+from langchain.embeddings import JinaEmbeddings
+from langchain.embeddings.huggingface import HuggingFaceEmbeddings
+from langchain.prompts import PromptTemplate
+from langchain.prompts.prompt import PromptTemplate
+from langchain.vectorstores import FAISS
+
+from chatllm import ChatLLM
+from chinese_text_splitter import ChineseTextSplitter
+
+nltk.data.path.append('./nltk_data')
+
+embedding_model_dict = {
+    "ernie-tiny": "nghuyong/ernie-3.0-nano-zh",
+    "ernie-base": "nghuyong/ernie-3.0-base-zh",
+    "text2vec-base": "GanymedeNil/text2vec-base-chinese",
+    "ViT-B-32": 'ViT-B-32::laion2b-s34b-b79k'
+}
+
+llm_model_dict = {
+    "ChatGLM-6B-int8": "THUDM/chatglm-6b-int8",
+    "ChatGLM-6B-int4": "THUDM/chatglm-6b-int4",
+    "ChatGLM-6b-int4-qe": "THUDM/chatglm-6b-int4-qe",
+    "Minimax": "Minimax"
+}
+
+DEVICE = "cuda" if torch.cuda.is_available(
+) else "mps" if torch.backends.mps.is_available() else "cpu"
+
+
+def search_web(query):
+
+    SESSION.proxies = {
+        "http": f"socks5h://localhost:7890",
+        "https": f"socks5h://localhost:7890"
+    }
+    results = ddg(query)
+    web_content = ''
+    if results:
+        for result in results:
+            web_content += result['body']
+    return web_content
+
+
+def load_file(filepath):
+    if filepath.lower().endswith(".pdf"):
+        loader = UnstructuredFileLoader(filepath)
+        textsplitter = ChineseTextSplitter(pdf=True)
+        docs = loader.load_and_split(textsplitter)
+    else:
+        loader = UnstructuredFileLoader(filepath, mode="elements")
+        textsplitter = ChineseTextSplitter(pdf=False)
+        docs = loader.load_and_split(text_splitter=textsplitter)
+    return docs
+
+
+def init_knowledge_vector_store(embedding_model, filepath):
+    if embedding_model == "ViT-B-32":
+        jina_auth_token = os.getenv('jina_auth_token')
+        embeddings = JinaEmbeddings(
+            jina_auth_token=jina_auth_token,
+            model_name=embedding_model_dict[embedding_model])
+    else:
+        embeddings = HuggingFaceEmbeddings(
+            model_name=embedding_model_dict[embedding_model], )
+        embeddings.client = sentence_transformers.SentenceTransformer(
+            embeddings.model_name, device=DEVICE)
+
+    docs = load_file(filepath)
+
+    vector_store = FAISS.from_documents(docs, embeddings)
+    return vector_store
+
+
+def get_knowledge_based_answer(query,
+                               large_language_model,
+                               vector_store,
+                               VECTOR_SEARCH_TOP_K,
+                               web_content,
+                               history_len,
+                               temperature,
+                               top_p,
+                               chat_history=[]):
+    if web_content:
+        prompt_template = f"""基于以下已知信息，简洁和专业的来回答用户的问题。
+                            如果无法从中得到答案，请说 "根据已知信息无法回答该问题" 或 "没有提供足够的相关信息"，不允许在答案中添加编造成分，答案请使用中文。
+                            已知网络检索内容：{web_content}""" + """
+                            已知内容:
+                            {context}
+                            问题:
+                            {question}"""
+    else:
+        prompt_template = """基于以下已知信息，请简洁并专业地回答用户的问题。
+            如果无法从中得到答案，请说 "根据已知信息无法回答该问题" 或 "没有提供足够的相关信息"。不允许在答案中添加编造成分。另外，答案请使用中文。
+
+            已知内容:
+            {context}
+
+            问题:
+            {question}"""
+    prompt = PromptTemplate(template=prompt_template,
+                            input_variables=["context", "question"])
+    chatLLM = ChatLLM()
+    chatLLM.history = chat_history[-history_len:] if history_len > 0 else []
+    if large_language_model == "Minimax":
+        chatLLM.model = 'Minimax'
+    else:
+        chatLLM.load_model(
+            model_name_or_path=llm_model_dict[large_language_model])
+        chatLLM.temperature = temperature
+        chatLLM.top_p = top_p
+
+    knowledge_chain = RetrievalQA.from_llm(
+        llm=chatLLM,
+        retriever=vector_store.as_retriever(
+            search_kwargs={"k": VECTOR_SEARCH_TOP_K}),
+        prompt=prompt)
+    knowledge_chain.combine_documents_chain.document_prompt = PromptTemplate(
+        input_variables=["page_content"], template="{page_content}")
+
+    knowledge_chain.return_source_documents = True
+
+    result = knowledge_chain({"query": query})
+    return result
+
+
+def clear_session():
+    return '', None
+
+
+def predict(input,
+            large_language_model,
+            embedding_model,
+            file_obj,
+            VECTOR_SEARCH_TOP_K,
+            history_len,
+            temperature,
+            top_p,
+            use_web,
+            history=None):
+    if history == None:
+        history = []
+    print(file_obj.name)
+    vector_store = init_knowledge_vector_store(embedding_model, file_obj.name)
+    if use_web == 'True':
+        web_content = search_web(query=input)
+    else:
+        web_content = ''
+    resp = get_knowledge_based_answer(
+        query=input,
+        large_language_model=large_language_model,
+        vector_store=vector_store,
+        VECTOR_SEARCH_TOP_K=VECTOR_SEARCH_TOP_K,
+        web_content=web_content,
+        chat_history=history,
+        history_len=history_len,
+        temperature=temperature,
+        top_p=top_p,
+    )
+    print(resp)
+    history.append((input, resp['result']))
+    return '', history, history
+
+
+if __name__ == "__main__":
+    block = gr.Blocks()
+    with block as demo:
+        gr.Markdown("""<h1><center>LangChain-ChatLLM-Webui</center></h1>
+        <center><font size=3>
+        本项目基于LangChain和大型语言模型系列模型, 提供基于本地知识的自动问答应用. <br>
+        目前项目提供基于<a href='https://github.com/THUDM/ChatGLM-6B' target="_blank">ChatGLM-6B </a>系列、Minimax的LLM和包括text2vec-base-chinese、ernie-3.0-zh系列以及由<a href='https://cloud.jina.ai/user/inference' target="_blank">Jina</a>提供的ViT-B-32::laion2b-s34b-b79k等多个Embedding模型, 支持上传 txt、docx、md等文本格式文件. <br>
+        后续将提供更加多样化的LLM、Embedding和参数选项供用户尝试, 欢迎关注<a href='https://github.com/thomas-yanxin/LangChain-ChatGLM-Webui' target="_blank">Github地址</a>. <br>
+        本项目已内置开发者自己的key，用户无需输入自己的相关key. <br>
+        当然，更推荐您点击右上角的<strong>Duplicate this Space</strong>，将项目Fork到自己的Space中，保护个人隐私，且避免排队！
+        </center></font>
+        """)
+        with gr.Row():
+            with gr.Column(scale=1):
+                model_choose = gr.Accordion("模型选择")
+                with model_choose:
+                    large_language_model = gr.Dropdown(
+                        list(llm_model_dict.keys()),
+                        label="large language model",
+                        value="ChatGLM-6B-int4")
+
+                    embedding_model = gr.Dropdown(list(
+                        embedding_model_dict.keys()),
+                                                  label="Embedding model",
+                                                  value="text2vec-base")
+
+                file = gr.File(label='请上传知识库文件, 目前支持txt、docx、md格式',
+                               file_types=['.txt', '.md', '.docx'])
+
+                use_web = gr.Radio(["True", "False"],
+                                   label="Web Search",
+                                   value="False")
+                model_argument = gr.Accordion("模型参数配置")
+
+                with model_argument:
+
+                    VECTOR_SEARCH_TOP_K = gr.Slider(
+                        1,
+                        10,
+                        value=6,
+                        step=1,
+                        label="vector search top k",
+                        interactive=True)
+
+                    HISTORY_LEN = gr.Slider(0,
+                                            3,
+                                            value=0,
+                                            step=1,
+                                            label="history len",
+                                            interactive=True)
+
+                    temperature = gr.Slider(0,
+                                            1,
+                                            value=0.01,
+                                            step=0.01,
+                                            label="temperature",
+                                            interactive=True)
+                    top_p = gr.Slider(0,
+                                      1,
+                                      value=0.9,
+                                      step=0.1,
+                                      label="top_p",
+                                      interactive=True)
+
+            with gr.Column(scale=4):
+                chatbot = gr.Chatbot(label='ChatLLM').style(height=600)
+                message = gr.Textbox(label='请输入问题')
+                state = gr.State()
+
+                with gr.Row():
+                    clear_history = gr.Button("🧹 清除历史对话")
+                    send = gr.Button("🚀 发送")
+
+                    send.click(predict,
+                               inputs=[
+                                   message, large_language_model,
+                                   embedding_model, file, VECTOR_SEARCH_TOP_K,
+                                   HISTORY_LEN, temperature, top_p, use_web,
+                                   state
+                               ],
+                               outputs=[message, chatbot, state])
+                    clear_history.click(fn=clear_session,
+                                        inputs=[],
+                                        outputs=[chatbot, state],
+                                        queue=False)
+
+                    message.submit(predict,
+                                   inputs=[
+                                       message, large_language_model,
+                                       embedding_model, file,
+                                       VECTOR_SEARCH_TOP_K, HISTORY_LEN,
+                                       temperature, top_p, use_web, state
+                                   ],
+                                   outputs=[message, chatbot, state])
+        gr.Markdown("""提醒：<br>
+        1. 使用时请先上传自己的知识文件，并且文件中不含某些特殊字符，否则将返回error. <br>
+        2. 有任何使用问题，请通过[问题交流区](https://huggingface.co/spaces/thomas-yanxin/LangChain-ChatLLM/discussions)或[Github Issue区](https://github.com/thomas-yanxin/LangChain-ChatGLM-Webui/issues)进行反馈. <br>
+        """)
+    demo.queue().launch(server_name='0.0.0.0', share=False)

chatllm.py

@@ -0,0 +1,159 @@
+
+import os
+from typing import Dict, List, Optional, Tuple, Union
+
+import torch
+from langchain.llms.base import LLM
+from langchain.llms.utils import enforce_stop_tokens
+from transformers import AutoModel, AutoTokenizer
+
+os.environ["TOKENIZERS_PARALLELISM"] = "false"
+
+DEVICE = "cuda"
+DEVICE_ID = "0"
+CUDA_DEVICE = f"{DEVICE}:{DEVICE_ID}" if DEVICE_ID else DEVICE
+
+
+def torch_gc():
+    if torch.cuda.is_available():
+        with torch.cuda.device(CUDA_DEVICE):
+            torch.cuda.empty_cache()
+            torch.cuda.ipc_collect()
+
+def auto_configure_device_map(num_gpus: int) -> Dict[str, int]:
+    # transformer.word_embeddings 占用1层
+    # transformer.final_layernorm 和 lm_head 占用1层
+    # transformer.layers 占用 28 层
+    # 总共30层分配到num_gpus张卡上
+    num_trans_layers = 28
+    per_gpu_layers = 30 / num_gpus
+
+    # bugfix: 在linux中调用torch.embedding传入的weight,input不在同一device上,导致RuntimeError
+    # windows下 model.device 会被设置成 transformer.word_embeddings.device
+    # linux下 model.device 会被设置成 lm_head.device
+    # 在调用chat或者stream_chat时,input_ids会被放到model.device上
+    # 如果transformer.word_embeddings.device和model.device不同,则会导致RuntimeError
+    # 因此这里将transformer.word_embeddings,transformer.final_layernorm,lm_head都放到第一张卡上
+    device_map = {'transformer.word_embeddings': 0,
+                  'transformer.final_layernorm': 0, 'lm_head': 0}
+
+    used = 2
+    gpu_target = 0
+    for i in range(num_trans_layers):
+        if used >= per_gpu_layers:
+            gpu_target += 1
+            used = 0
+        assert gpu_target < num_gpus
+        device_map[f'transformer.layers.{i}'] = gpu_target
+        used += 1
+
+    return device_map
+
+
+
+class ChatLLM(LLM):
+    max_token: int = 10000
+    temperature: float = 0.1
+    top_p = 0.9
+    history = []
+    tokenizer: object = None
+    model: object = None
+
+    def __init__(self):
+        super().__init__()
+
+    @property
+    def _llm_type(self) -> str:
+        return "ChatLLM"
+
+    def _call(self,
+              prompt: str,
+              stop: Optional[List[str]] = None) -> str:
+        
+        if self.model == 'Minimax':
+            import requests
+
+            group_id = os.getenv('group_id')
+            api_key = os.getenv('api_key')
+
+            url = f'https://api.minimax.chat/v1/text/chatcompletion?GroupId={group_id}'
+            headers = {
+                "Authorization": f"Bearer {api_key}",
+                "Content-Type": "application/json"
+            }
+            request_body = {
+                "model": "abab5-chat",
+                "tokens_to_generate": 512,
+                'messages': []
+            }
+
+            for i in self.history:
+                h_input = i[0]
+                h_reply = i[1]
+                request_body['messages'].append({
+                    "sender_type": "USER",
+                    "text": h_input
+                })
+                request_body['messages'].append({"sender_type": "BOT", "text": h_reply})
+
+            request_body['messages'].append({"sender_type": "USER", "text": prompt})
+            resp = requests.post(url, headers=headers, json=request_body)
+            response = resp.json()['reply']
+            #  将当次的ai回复内容加入messages
+            request_body['messages'].append({"sender_type": "BOT", "text": response})
+            self.history.append((prompt, response))
+        
+        else:
+
+            response, _ = self.model.chat(
+                self.tokenizer,
+                prompt,
+                history=self.history,
+                max_length=self.max_token,
+                temperature=self.temperature,
+            )
+            torch_gc()
+            if stop is not None:
+                response = enforce_stop_tokens(response, stop)
+            self.history = self.history+[[None, response]]
+        return response
+
+    def load_model(self,
+                   model_name_or_path: str = "THUDM/chatglm-6b-int4",
+                   llm_device=DEVICE,
+                   device_map: Optional[Dict[str, int]] = None,
+                   **kwargs):
+        self.tokenizer = AutoTokenizer.from_pretrained(
+            model_name_or_path,
+            trust_remote_code=True
+        )
+        if torch.cuda.is_available() and llm_device.lower().startswith("cuda"):
+            # 根据当前设备GPU数量决定是否进行多卡部署
+            num_gpus = torch.cuda.device_count()
+            if num_gpus < 2 and device_map is None:
+                self.model = (
+                    AutoModel.from_pretrained(
+                        model_name_or_path, 
+                        trust_remote_code=True, 
+                        **kwargs)
+                    .half()
+                    .cuda()
+                )
+            else:
+                from accelerate import dispatch_model
+
+                model = AutoModel.from_pretrained(model_name_or_path, trust_remote_code=True, **kwargs).half()
+                # 可传入device_map自定义每张卡的部署情况
+                if device_map is None:
+                    device_map = auto_configure_device_map(num_gpus)
+
+                self.model = dispatch_model(model, device_map=device_map)
+        else:
+            self.model = (
+                AutoModel.from_pretrained(
+                    model_name_or_path,
+                    trust_remote_code=True)
+                .float()
+                .to(llm_device)
+            )
+        self.model = self.model.eval()

chinese_text_splitter.py

@@ -0,0 +1,24 @@
+import re
+from typing import List
+
+from langchain.text_splitter import CharacterTextSplitter
+
+
+class ChineseTextSplitter(CharacterTextSplitter):
+    def __init__(self, pdf: bool = False, **kwargs):
+        super().__init__(**kwargs)
+        self.pdf = pdf
+
+    def split_text(self, text: str) -> List[str]:
+        if self.pdf:
+            text = re.sub(r"\n{3,}", "\n", text)
+            text = re.sub('\s', ' ', text)
+            text = text.replace("\n\n", "")
+        sent_sep_pattern = re.compile('([﹒﹔﹖﹗．。！？]["’”」』]{0,2}|(?=["‘“「『]{1,2}|$))')  # del ：；
+        sent_list = []
+        for ele in sent_sep_pattern.split(text):
+            if sent_sep_pattern.match(ele) and sent_list:
+                sent_list[-1] += ele
+            elif ele:
+                sent_list.append(ele)
+        return sent_list

nltk_data/taggers/averaged_perceptron_tagger/averaged_perceptron_tagger.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:25a5a19c7ced7b2bac3831da5bc0afcc2c34e5dd01cd4f361bb799949a696238
+size 6138625

nltk_data/tokenizers/punkt/PY3/README

@@ -0,0 +1,98 @@
+Pretrained Punkt Models -- Jan Strunk (New version trained after issues 313 and 514 had been corrected)
+
+Most models were prepared using the test corpora from Kiss and Strunk (2006). Additional models have
+been contributed by various people using NLTK for sentence boundary detection.
+
+For information about how to use these models, please confer the tokenization HOWTO:
+http://nltk.googlecode.com/svn/trunk/doc/howto/tokenize.html
+and chapter 3.8 of the NLTK book:
+http://nltk.googlecode.com/svn/trunk/doc/book/ch03.html#sec-segmentation
+
+There are pretrained tokenizers for the following languages:
+
+File                Language            Source                             Contents                Size of training corpus(in tokens)           Model contributed by
+=======================================================================================================================================================================
+czech.pickle        Czech               Multilingual Corpus 1 (ECI)        Lidove Noviny                   ~345,000                             Jan Strunk / Tibor Kiss
+                                                                           Literarni Noviny
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+danish.pickle       Danish              Avisdata CD-Rom Ver. 1.1. 1995     Berlingske Tidende              ~550,000                             Jan Strunk / Tibor Kiss
+                                        (Berlingske Avisdata, Copenhagen)  Weekend Avisen
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+dutch.pickle        Dutch               Multilingual Corpus 1 (ECI)        De Limburger                    ~340,000                             Jan Strunk / Tibor Kiss
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+english.pickle      English             Penn Treebank (LDC)                Wall Street Journal             ~469,000                             Jan Strunk / Tibor Kiss
+                    (American)
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+estonian.pickle     Estonian            University of Tartu, Estonia       Eesti Ekspress                  ~359,000                             Jan Strunk / Tibor Kiss
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+finnish.pickle      Finnish             Finnish Parole Corpus, Finnish     Books and major national        ~364,000                             Jan Strunk / Tibor Kiss
+                                        Text Bank (Suomen Kielen           newspapers
+                                        Tekstipankki)
+                                        Finnish Center for IT Science
+                                        (CSC)
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+french.pickle       French              Multilingual Corpus 1 (ECI)        Le Monde                        ~370,000                             Jan Strunk / Tibor Kiss
+                    (European)
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+german.pickle       German              Neue Zürcher Zeitung AG            Neue Zürcher Zeitung            ~847,000                             Jan Strunk / Tibor Kiss
+                    (Switzerland)       CD-ROM
+                    (Uses "ss"
+                     instead of "ß")
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+greek.pickle        Greek               Efstathios Stamatatos              To Vima (TO BHMA)               ~227,000                             Jan Strunk / Tibor Kiss
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+italian.pickle      Italian             Multilingual Corpus 1 (ECI)        La Stampa, Il Mattino           ~312,000                             Jan Strunk / Tibor Kiss
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+norwegian.pickle    Norwegian           Centre for Humanities              Bergens Tidende                 ~479,000                             Jan Strunk / Tibor Kiss
+                    (Bokmål and         Information Technologies,
+                     Nynorsk)           Bergen
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+polish.pickle       Polish              Polish National Corpus             Literature, newspapers, etc.  ~1,000,000                             Krzysztof Langner
+                                        (http://www.nkjp.pl/)
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+portuguese.pickle   Portuguese          CETENFolha Corpus                  Folha de São Paulo              ~321,000                             Jan Strunk / Tibor Kiss
+                    (Brazilian)         (Linguateca)
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+slovene.pickle      Slovene             TRACTOR                            Delo                            ~354,000                             Jan Strunk / Tibor Kiss
+                                        Slovene Academy for Arts
+                                        and Sciences
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+spanish.pickle      Spanish             Multilingual Corpus 1 (ECI)        Sur                             ~353,000                             Jan Strunk / Tibor Kiss
+                    (European)
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+swedish.pickle      Swedish             Multilingual Corpus 1 (ECI)        Dagens Nyheter                  ~339,000                             Jan Strunk / Tibor Kiss
+                                                                           (and some other texts)
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+turkish.pickle      Turkish             METU Turkish Corpus                Milliyet                        ~333,000                             Jan Strunk / Tibor Kiss
+                                        (Türkçe Derlem Projesi)
+                                        University of Ankara
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+
+The corpora contained about 400,000 tokens on average and mostly consisted of newspaper text converted to
+Unicode using the codecs module.
+
+Kiss, Tibor and Strunk, Jan (2006): Unsupervised Multilingual Sentence Boundary Detection.
+Computational Linguistics 32: 485-525.
+
+---- Training Code ----
+
+# import punkt
+import nltk.tokenize.punkt
+
+# Make a new Tokenizer
+tokenizer = nltk.tokenize.punkt.PunktSentenceTokenizer()
+
+# Read in training corpus (one example: Slovene)
+import codecs
+text = codecs.open("slovene.plain","Ur","iso-8859-2").read()
+
+# Train tokenizer
+tokenizer.train(text)
+
+# Dump pickled tokenizer
+import pickle
+out = open("slovene.pickle","wb")
+pickle.dump(tokenizer, out)
+out.close()
+
+---------

nltk_data/tokenizers/punkt/PY3/czech.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64b0734b6fbe8e8d7cac79f48d1dd9f853824e57c4e3594dadd74ba2c1d97f50
+size 1119050

nltk_data/tokenizers/punkt/PY3/danish.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6189c7dd254e29e2bd406a7f6a4336297c8953214792466a790ea4444223ceb3
+size 1191710

nltk_data/tokenizers/punkt/PY3/dutch.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fda0d6a13f02e8898daec7fe923da88e25abe081bcfa755c0e015075c215fe4c
+size 693759

nltk_data/tokenizers/punkt/PY3/english.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5cad3758596392364e3be9803dbd7ebeda384b68937b488a01365f5551bb942c
+size 406697

nltk_data/tokenizers/punkt/PY3/estonian.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b364f72538d17b146a98009ad239a8096ce6c0a8b02958c0bc776ecd0c58a25f
+size 1499502

nltk_data/tokenizers/punkt/PY3/finnish.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a4b5ff5500ee851c456f9dd40d5fc0d8c1859c88eb3178de1317d26b7d22833
+size 1852226

nltk_data/tokenizers/punkt/PY3/french.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28e3a4cd2971989b3cb9fd3433a6f15d17981e464db2be039364313b5de94f29
+size 553575

nltk_data/tokenizers/punkt/PY3/german.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddcbbe85e2042a019b1a6e37fd8c153286c38ba201fae0f5bfd9a3f74abae25c
+size 1463575

nltk_data/tokenizers/punkt/PY3/greek.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:85dabc44ab90a5f208ef37ff6b4892ebe7e740f71fb4da47cfd95417ca3e22fd
+size 876006

nltk_data/tokenizers/punkt/PY3/italian.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:68a94007b1e4ffdc4d1a190185ca5442c3dafeb17ab39d30329e84cd74a43947
+size 615089

nltk_data/tokenizers/punkt/PY3/malayalam.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f8cf58acbdb7f472ac40affc13663be42dafb47c15030c11ade0444c9e0e53d
+size 221207

nltk_data/tokenizers/punkt/PY3/norwegian.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ff7a46d1438b311457d15d7763060b8d3270852c1850fd788c5cee194dc4a1d
+size 1181271

nltk_data/tokenizers/punkt/PY3/polish.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:624900ae3ddfb4854a98c5d3b8b1c9bb719975f33fee61ce1441dab9f8a00718
+size 1738386

nltk_data/tokenizers/punkt/PY3/portuguese.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02a0b7b25c3c7471e1791b66a31bbb530afbb0160aee4fcecf0107652067b4a1
+size 611919

nltk_data/tokenizers/punkt/PY3/russian.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:549762f8190024d89b511472df21a3a135eee5d9233e63ac244db737c2c61d7e
+size 33020

nltk_data/tokenizers/punkt/PY3/slovene.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52ef2cc0ed27d79b3aa635cbbc40ad811883a75a4b8a8be1ae406972870fd864
+size 734444

nltk_data/tokenizers/punkt/PY3/spanish.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:164a50fadc5a49f8ec7426eae11d3111ee752b48a3ef373d47745011192a5984
+size 562337

nltk_data/tokenizers/punkt/PY3/swedish.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0f7d538bfd5266633b09e842cd92e9e0ac10f1d923bf211e1497972ddc47318
+size 979681

nltk_data/tokenizers/punkt/PY3/turkish.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae68ef5863728ac5332e87eb1f6bae772ff32a13a4caa2b01a5c68103e853c5b
+size 1017038

nltk_data/tokenizers/punkt/README

@@ -0,0 +1,98 @@
+Pretrained Punkt Models -- Jan Strunk (New version trained after issues 313 and 514 had been corrected)
+
+Most models were prepared using the test corpora from Kiss and Strunk (2006). Additional models have
+been contributed by various people using NLTK for sentence boundary detection.
+
+For information about how to use these models, please confer the tokenization HOWTO:
+http://nltk.googlecode.com/svn/trunk/doc/howto/tokenize.html
+and chapter 3.8 of the NLTK book:
+http://nltk.googlecode.com/svn/trunk/doc/book/ch03.html#sec-segmentation
+
+There are pretrained tokenizers for the following languages:
+
+File                Language            Source                             Contents                Size of training corpus(in tokens)           Model contributed by
+=======================================================================================================================================================================
+czech.pickle        Czech               Multilingual Corpus 1 (ECI)        Lidove Noviny                   ~345,000                             Jan Strunk / Tibor Kiss
+                                                                           Literarni Noviny
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+danish.pickle       Danish              Avisdata CD-Rom Ver. 1.1. 1995     Berlingske Tidende              ~550,000                             Jan Strunk / Tibor Kiss
+                                        (Berlingske Avisdata, Copenhagen)  Weekend Avisen
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+dutch.pickle        Dutch               Multilingual Corpus 1 (ECI)        De Limburger                    ~340,000                             Jan Strunk / Tibor Kiss
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+english.pickle      English             Penn Treebank (LDC)                Wall Street Journal             ~469,000                             Jan Strunk / Tibor Kiss
+                    (American)
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+estonian.pickle     Estonian            University of Tartu, Estonia       Eesti Ekspress                  ~359,000                             Jan Strunk / Tibor Kiss
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+finnish.pickle      Finnish             Finnish Parole Corpus, Finnish     Books and major national        ~364,000                             Jan Strunk / Tibor Kiss
+                                        Text Bank (Suomen Kielen           newspapers
+                                        Tekstipankki)
+                                        Finnish Center for IT Science
+                                        (CSC)
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+french.pickle       French              Multilingual Corpus 1 (ECI)        Le Monde                        ~370,000                             Jan Strunk / Tibor Kiss
+                    (European)
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+german.pickle       German              Neue Zürcher Zeitung AG            Neue Zürcher Zeitung            ~847,000                             Jan Strunk / Tibor Kiss
+                    (Switzerland)       CD-ROM
+                    (Uses "ss"
+                     instead of "ß")
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+greek.pickle        Greek               Efstathios Stamatatos              To Vima (TO BHMA)               ~227,000                             Jan Strunk / Tibor Kiss
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+italian.pickle      Italian             Multilingual Corpus 1 (ECI)        La Stampa, Il Mattino           ~312,000                             Jan Strunk / Tibor Kiss
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+norwegian.pickle    Norwegian           Centre for Humanities              Bergens Tidende                 ~479,000                             Jan Strunk / Tibor Kiss
+                    (Bokmål and         Information Technologies,
+                     Nynorsk)           Bergen
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+polish.pickle       Polish              Polish National Corpus             Literature, newspapers, etc.  ~1,000,000                             Krzysztof Langner
+                                        (http://www.nkjp.pl/)
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+portuguese.pickle   Portuguese          CETENFolha Corpus                  Folha de São Paulo              ~321,000                             Jan Strunk / Tibor Kiss
+                    (Brazilian)         (Linguateca)
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+slovene.pickle      Slovene             TRACTOR                            Delo                            ~354,000                             Jan Strunk / Tibor Kiss
+                                        Slovene Academy for Arts
+                                        and Sciences
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+spanish.pickle      Spanish             Multilingual Corpus 1 (ECI)        Sur                             ~353,000                             Jan Strunk / Tibor Kiss
+                    (European)
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+swedish.pickle      Swedish             Multilingual Corpus 1 (ECI)        Dagens Nyheter                  ~339,000                             Jan Strunk / Tibor Kiss
+                                                                           (and some other texts)
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+turkish.pickle      Turkish             METU Turkish Corpus                Milliyet                        ~333,000                             Jan Strunk / Tibor Kiss
+                                        (Türkçe Derlem Projesi)
+                                        University of Ankara
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------
+
+The corpora contained about 400,000 tokens on average and mostly consisted of newspaper text converted to
+Unicode using the codecs module.
+
+Kiss, Tibor and Strunk, Jan (2006): Unsupervised Multilingual Sentence Boundary Detection.
+Computational Linguistics 32: 485-525.
+
+---- Training Code ----
+
+# import punkt
+import nltk.tokenize.punkt
+
+# Make a new Tokenizer
+tokenizer = nltk.tokenize.punkt.PunktSentenceTokenizer()
+
+# Read in training corpus (one example: Slovene)
+import codecs
+text = codecs.open("slovene.plain","Ur","iso-8859-2").read()
+
+# Train tokenizer
+tokenizer.train(text)
+
+# Dump pickled tokenizer
+import pickle
+out = open("slovene.pickle","wb")
+pickle.dump(tokenizer, out)
+out.close()
+
+---------

nltk_data/tokenizers/punkt/czech.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c085f6283bed0f1390d36a55d126ccc29c9b4dfcd2705e862b1711b7c6bb5ab
+size 1424691

nltk_data/tokenizers/punkt/danish.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df8366ad67db22b1f838cd63fcc589a6006faf66d7a46be5312d9c487ce2c811
+size 1427491

nltk_data/tokenizers/punkt/dutch.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12f46024d3c840529b56ac2a3118b80b8dc77705734bcdd71ff7c46f5808395e
+size 839761

nltk_data/tokenizers/punkt/english.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e2d25d5adc3ee51ac192ce611bdc5378acae7136af5d3c52c2903c669f9aff0
+size 495006

nltk_data/tokenizers/punkt/estonian.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9083ef6ef3d5b9992a8a4ea09e889a87be75e2122ad25648307178960634cd8d
+size 1803082

nltk_data/tokenizers/punkt/finnish.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce1b4dbe72e400e902220061457f9bd5f491ec37f7af468bc4694980c9623817
+size 2192034

nltk_data/tokenizers/punkt/french.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e0be48e38a54232ea88c817cf34c1f1f8f44954e21f118c65af9f2d6a43cdbd
+size 664010

nltk_data/tokenizers/punkt/german.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:251c2f4bde61ab3fc1cabc2158c62e6ab285fddd16267d2d3885f71e3ed61c7f
+size 1708012

nltk_data/tokenizers/punkt/greek.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b3a6da093ed2df084ded6dc49c88f101d47a0c69398f19ae50af6785d93b1c5
+size 2042362

nltk_data/tokenizers/punkt/italian.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41e6aaf554e696703b3d41890973368b9b2f17c342745c07369742928d363731
+size 748532

nltk_data/tokenizers/punkt/malayalam.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f8cf58acbdb7f472ac40affc13663be42dafb47c15030c11ade0444c9e0e53d
+size 221207

nltk_data/tokenizers/punkt/norwegian.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:45828b0d57da9a66f107ea277752f6c1cbde51b9f9feba173b2c6e2edb28af21
+size 1422756

nltk_data/tokenizers/punkt/polish.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79d09a9406f90dbf20f8cbb0a04a7aa0bdb4b71604eda31e97c3df2de5cd2837
+size 2287622

nltk_data/tokenizers/punkt/portuguese.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c09561e770b6f17e3d85112f83007ff1397dec66c23acb15b9fe046eaefd2e86
+size 739845

nltk_data/tokenizers/punkt/russian.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc984432fbe31f7000014f8047502476889169c60f09be5413ca09276b16c909
+size 33027

nltk_data/tokenizers/punkt/slovene.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2dc83b900e347c16ed0123868369107cd19d1a6125d099e26889580c4dbba277
+size 939791

nltk_data/tokenizers/punkt/spanish.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61afae663cb2968148e0e27d5a3fcd4a5f19648688800caf8e7f998eaa75f4a7
+size 680466

nltk_data/tokenizers/punkt/swedish.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e5794208b223b2a54bd4ed565045172f9c6ef80b5bead94f71a5499455cda955
+size 1168214

nltk_data/tokenizers/punkt/turkish.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2abb5d7ec4e80aeeb994407254a2e1a0928520727cc25f7bd3fc9ce0b5a78c1
+size 1363199

requirements.txt

@@ -0,0 +1,19 @@
+torch
+langchain
+accelerate
+duckduckgo_search
+transformers==4.27.1
+unstructured[local-inference]
+layoutparser[layoutmodels,tesseract]
+nltk
+sentence-transformers
+beautifulsoup4
+icetk
+cpm_kernels
+faiss-cpu
+gradio
+nltk
+torch
+torchvision
+protobuf==3.19
+jina