watsonx.aiのllama3でchatbot

LCELを使えばOK。

# %%
from dotenv import load_dotenv
import os

load_dotenv()
#print(os.environ['WATSONX_APIKEY'])
# %%
from langchain.globals import set_verbose

set_verbose(True)
# %%
from langchain_ibm.llms import WatsonxLLM

model_id = 'meta-llama/llama-3-70b-instruct'
project_id = '13facb99-f76c-4679-94c5-14e692b114db'
url = 'https://jp-tok.ml.cloud.ibm.com'
params = {
	'decoding_method': 'sample',
	'max_new_tokens': 4096,
	'top_k': 50,
	'top_p': 1,
	'repetition_penalty': 1
}
llm = WatsonxLLM(
    model_id=model_id,
    project_id=project_id,
    url=url,
    params=params
)
# %%
from langchain_core.prompts.prompt import PromptTemplate

template = """
<|begin_of_text|>
<|start_header_id|>system<|end_header_id|>You are a helpful, respectful and honest assistant. Always answer as helpfully as possible, while being safe. Your answers should not include any harmful, unethical, racist, sexist, toxic, dangerous, or illegal content. Please ensure that your responses are socially unbiased and positive in nature. If a question does not make any sense, or is not factually coherent, explain why instead of answering something not correct. If you don’t know the answer to a question, please don’t share false information.<|eot_id|>
<|start_header_id|>user<|end_header_id|>
{history}
Human: {user}
<|eot_id|>
<|start_header_id|>assistant<|end_header_id|>
"""
prompt = PromptTemplate(template=template, input_variables=['history', 'user'])
# %%
from langchain.memory.buffer_window import ConversationBufferWindowMemory

memory = ConversationBufferWindowMemory()
# %%
from langchain_core.runnables.passthrough import RunnablePassthrough
from langchain_core.output_parsers import StrOutputParser

chain = RunnablePassthrough().assign(
    history = lambda x: memory.chat_memory
) | prompt | llm | StrOutputParser()
# %%
#user = 'watsonx.aiとは何ですか？'
#ai = ''
#for tmp in chain.stream({'user': user}):
#    ai += tmp
#    print(tmp)
#memory.chat_memory.add_user_message(user)
#memory.chat_memory.add_ai_message(ai)
#print(memory.chat_memory)
# %%

import gradio as gr
from llama3 import chain, memory

def fn(message, history):
    user = message
    ai = ''
    for tmp in chain.stream({'user': user}):
        ai += tmp
        yield ai
    memory.chat_memory.add_user_message(user)
    memory.chat_memory.add_ai_message(ai)
    #print(memory.chat_memory)

gr.ChatInterface(fn=fn).launch()