towerinstruct translate English to French

towerinstruct_lm_studio_client.py

#!/usr/bin/env python3

import json
import re
import sys
from pathlib import Path

# install with: pip3 install openai — see https://pypi.org/project/openai/
from openai import OpenAI

client = OpenAI(base_url="http://localhost:1234/v1")
MODEL = "towerinstruct-13b-v0.1"

input_file = Path(sys.argv[1])

if not input_file.exists():
    print(f'{input_file} not found!')
    exit(1)

output = input_file.with_suffix(f'._FR_{MODEL}_LM_STUDIO.html')
output_JSON = input_file.with_suffix(f'._FR_{MODEL}_LM_STUDIO.json')

if not output_JSON.exists():
    with output_JSON.open(mode='w', encoding='utf-8') as jwr:
        json.dump([], fp=jwr, ensure_ascii=False)

PATTERN_PAR = re.compile("(?=<p>|<h3>)")


source_text = input_file.read_text(encoding='utf-8')
fragments = PATTERN_PAR.split(source_text)

for fragment in fragments:
    response = client.chat.completions.create(
        model=MODEL,
        messages=[
            {
                "role": "user",
                "content": f"""Translate English to French:
    English:
    {fragment}
    """,
            },
        ],
        stream=False,
        temperature=0.2,
        max_tokens=-1
    )

    content = response.choices[0].message.content
    print(f'🇬🇧  {fragment}')
    print(f"🇫🇷  {content}")

    with output.open("a", encoding="utf-8") as fout:
      fout.write(f"{content}\n")

    with output_JSON.open(mode='r', encoding='utf-8') as jr:
        data = json.load(fp=jr) or []
        
    with output_JSON.open(mode="w", encoding="utf-8") as jwr:
        data.append([fragment.strip(), content.strip()])
        json.dump(data, fp=jwr, indent=1, ensure_ascii=False)