摘要: # 优化TF ## TF衰减 > 思考一个问题:一篇文档,里面有提到200次RedCap,一定是2倍相关于另一篇提到100次RedCap吗? $TF$对匹配度的贡献应该是有所衰减的。如何控制衰减曲线的陡峭程度?构造$TF$衰减的匹配度函数的一个trick是引入参数$k$: $$ {TF}^\prim 阅读全文
posted @ 2023-08-18 16:22 LexLuc 阅读(597) 评论(0) 推荐(0)
摘要: ```bash sudo apt-get install openssl cd /etc/nginx sudo mkdir ssl sudo openssl req -batch -x509 -nodes -days 365 \ -newkey rsa:2048 \ -keyout /etc/ngi 阅读全文
posted @ 2023-07-18 02:00 LexLuc 阅读(29) 评论(0) 推荐(0)
摘要: ```python import time from py2neo import Graph from config.neo4j_config import NEO4J_CONFIG_PRO # NEO4J_CONFIG_PRO = { # "host": "x.x.x.x", # "port": 阅读全文
posted @ 2023-07-11 22:20 LexLuc 阅读(98) 评论(0) 推荐(0)
摘要: ```python # -*-coding:utf-8 -*- import redis from config.redis_config import CACHE_REDIS_CONF # CACHE_REDIS_CONF = { # "host": "x.x.x.x", # "port": ?? 阅读全文
posted @ 2023-07-11 22:17 LexLuc 阅读(20) 评论(0) 推荐(0)
摘要: ```python class MySqlClient: def __init__(self, db_name: str): self._conn = pymysql.connect(host=MYSQL_CONFIG['host'], port=MYSQL_CONFIG['port'], user 阅读全文
posted @ 2023-07-11 22:12 LexLuc 阅读(19) 评论(0) 推荐(0)
摘要: ``` # requirement.txt happybase==1.2.0 ``` ```python """ hbase.py """ # -*- coding: utf-8 -*- import happybase from config.hbase_config import HAPPYBA 阅读全文
posted @ 2023-07-11 22:07 LexLuc 阅读(24) 评论(0) 推荐(0)
摘要: ```python # -*- coding: utf-8 -*- import os import sys import shutil from time import perf_counter from collections import defaultdict from datetime i 阅读全文
posted @ 2023-07-11 21:41 LexLuc 阅读(18) 评论(0) 推荐(0)
摘要: # BASE模型 模型大小:15.5B 训练数据量:800GB (1T token) 硬件资源:512张 Tesla A100 训练时长:24天 # 微调模型 微调模型:ZeRO-3 硬件资源:8张 Tesla A100 训练时长:3小时 # 有用链接 https://github.com/bigc 阅读全文
posted @ 2023-07-08 21:00 LexLuc 阅读(807) 评论(0) 推荐(0)
摘要: ``` # Byte-compiled / optimized / DLL files __pycache__/ *.py[cod] *$py.class # C extensions *.so # Distribution / packaging .Python build/ develop-eg 阅读全文
posted @ 2023-07-04 00:50 LexLuc 阅读(35) 评论(0) 推荐(0)
摘要: # StarCoderBase HF: https://huggingface.co/bigcode/starcoderbase Training dataset: [The Stack v1.2](https://huggingface.co/datasets/bigcode/the-stack) 阅读全文
posted @ 2023-07-03 16:54 LexLuc 阅读(112) 评论(0) 推荐(0)