{}
".format(line.strip()) html_content += "" return html_content @spaces.GPU(duration=120) def deidentify_doc(pdftext, maxtokens, temperature, top_probability): # prompt = "In the following text replace any person name and any address with term [redacted], replace any Date of Birth and NHS number with term [redacted]" prompt = """ Perform the following actions on given report: 1. Replace any person names, age, date of birth, gender with term [redacted] 2. Replace any addresses with term [redacted] 3. DO NOT REPLACE ANY MEDICAL MEASUREMENTS 4. Replace only the CALENDAR DATES of format 'day/month/year' with term [redacted] """ # model_id = "models/Meta-Llama-3-70B-Instruct-Q3_K_M.gguf" # # model = Llama(model_path=model_id, n_ctx=2048, n_threads=8, n_gpu_layers=-1, n_batch=128) # model = Llama( # model_path=model_id, # flash_attn=True, # n_gpu_layers=81, # n_batch=1024, # n_ctx=8192, # ) chat_template = MessagesFormatterType.LLAMA_3 llm = Llama( model_path="models/Meta-Llama-3-70B-Instruct-Q3_K_M.gguf", flash_attn=True, n_gpu_layers=81, n_batch=1024, n_ctx=8192, ) provider = LlamaCppPythonProvider(llm) agent = LlamaCppAgent( provider, system_prompt="You are a helpful assistant.", predefined_messages_formatter_type=chat_template, debug_output=True ) settings = provider.get_provider_default_settings() settings.temperature = 0.7 settings.top_k = 40 settings.top_p = 0.95 settings.max_tokens = 2048 settings.repeat_penalty = 1.1 settings.stream = True messages = BasicChatHistory() stream = agent.get_chat_response( prompt + ' : ' + pdftext, llm_sampling_settings=settings, chat_history=messages, returns_streaming_generator=True, print_output=False ) outputs = "" for output in stream: outputs += output return outputs # output = model.create_chat_completion( # messages=[ # {"role": "assistant", "content": prompt}, # { # "role": "user", # "content": pdftext # } # ], # max_tokens=maxtokens, # temperature=temperature # ) # output = output['choices'][0]['message']['content'] # prompt = "Perform the following actions on given text: 1. Replace any person age with term [redacted] 2. DO NOT REPLACE ANY MEDICAL MEASUREMENTS 3. Replace only the CALENDAR DATES of format 'day/month/year' with term [redacted]" # output = model.create_chat_completion( # messages=[ # {"role": "assistant", "content": prompt}, # { # "role": "user", # "content": output # } # ], # max_tokens=maxtokens, # temperature=temperature # ) # output = output['choices'][0]['message']['content'] # print(prompt) # print(output) # print('-------------------------------------------------------') # return outputs def pdf_to_text(files, maxtokens=2048, temperature=0, top_probability=0.95): print('Control 0-----------------------------------') files=[files]#remove later for file in files: file_name = os.path.basename(file) file_name_splt = file_name.split('.') # print('File name is ', file_name) if (len(file_name_splt) > 1 and file_name_splt[1] == 'pdf'): page2content = process_document(file, page_ids=[0]) pdftext = page2content[1] print(pdftext) # pdftext = file # remove later if (pdftext): #shift this if block to right later anonymized_text = deidentify_doc(pdftext, maxtokens, temperature, top_probability) return anonymized_text # css = ".gradio-container {background: 'logo.png'}" # temp_slider = gr.Slider(minimum=0, maximum=2, value=0.9, label="Temperature Value") # prob_slider = gr.Slider(minimum=0, maximum=1, value=0.95, label="Max Probability Value") # max_tokens = gr.Number(value=600, label="Max Tokens") # input_folder = gr.File(file_count='multiple') # input_folder_text = gr.Textbox(label='Enter output folder path') # output_text = gr.Textbox() # output_path_component = gr.File(label="Select Output Path") # iface = gr.Interface( # fn=pdf_to_text, # inputs='file', # # inputs=["textbox", input_folder_text, "textbox", max_tokens, temp_slider, prob_slider], # outputs=output_text, # title='COBIx Endoscopy Report De-Identification', # description="This application assists to remove personal information from the uploaded clinical report", # theme=gr.themes.Soft(), # ) # iface.launch() # import spaces # import json # import subprocess # from llama_cpp import Llama # from llama_cpp_agent import LlamaCppAgent, MessagesFormatterType # from llama_cpp_agent.providers import LlamaCppPythonProvider # from llama_cpp_agent.chat_history import BasicChatHistory # from llama_cpp_agent.chat_history.messages import Roles # import gradio as gr # from huggingface_hub import hf_hub_download # hf_hub_download( # repo_id="bartowski/Meta-Llama-3-70B-Instruct-GGUF", # filename="Meta-Llama-3-70B-Instruct-Q3_K_M.gguf", # local_dir = "./models" # ) # # hf_hub_download( # # repo_id="bartowski/Mistral-7B-Instruct-v0.3-GGUF", # # filename="Mistral-7B-Instruct-v0.3-f32.gguf", # # local_dir = "./models" # # ) # css = """ # .message-row { # justify-content: space-evenly !important; # } # .message-bubble-border { # border-radius: 6px !important; # } # .message-buttons-bot, .message-buttons-user { # right: 10px !important; # left: auto !important; # bottom: 2px !important; # } # .dark.message-bubble-border { # border-color: #343140 !important; # } # .dark.user { # background: #1e1c26 !important; # } # .dark.assistant.dark, .dark.pending.dark { # background: #16141c !important; # } # """ # def get_messages_formatter_type(model_name): # if "Llama" in model_name: # return MessagesFormatterType.LLAMA_3 # elif "Mistral" in model_name: # return MessagesFormatterType.MISTRAL # else: # raise ValueError(f"Unsupported model: {model_name}") # @spaces.GPU(duration=60) # def respond( # message, # history: list[tuple[str, str]], # model, # system_message, # max_tokens, # temperature, # top_p, # top_k, # repeat_penalty, # ): # chat_template = get_messages_formatter_type(model) # llm = Llama( # model_path=f"models/{model}", # flash_attn=True, # n_gpu_layers=81, # n_batch=1024, # n_ctx=8192, # ) # provider = LlamaCppPythonProvider(llm) # agent = LlamaCppAgent( # provider, # system_prompt=f"{system_message}", # predefined_messages_formatter_type=chat_template, # debug_output=True # ) # settings = provider.get_provider_default_settings() # settings.temperature = temperature # settings.top_k = top_k # settings.top_p = top_p # settings.max_tokens = max_tokens # settings.repeat_penalty = repeat_penalty # settings.stream = True # messages = BasicChatHistory() # for msn in history: # user = { # 'role': Roles.user, # 'content': msn[0] # } # assistant = { # 'role': Roles.assistant, # 'content': msn[1] # } # messages.add_message(user) # messages.add_message(assistant) # stream = agent.get_chat_response( # message, # llm_sampling_settings=settings, # chat_history=messages, # returns_streaming_generator=True, # print_output=False # ) # outputs = "" # for output in stream: # outputs += output # yield outputs # PLACEHOLDER = """ #
# The llama-cpp-agent framework simplifies interactions with Large Language Models (LLMs), providing an interface for chatting, executing function calls, generating structured output, performing retrieval augmented generation, and processing text using agentic chains with tools.
# #