| ~ |

起点是 ChFinAnn 数据集, 和它的 SOTA 榜单. 其中

看起来很不错, 模型比其他人轻量很多, 训练资源要求少 (对比 Doc2EDAG 等涉及到动态建立图神经网络的巨大计算资源需求), 推理速度也快, 效果和其他人相当.

Read more »

| ~ | 
import urllib.parse

import sqlalchemy

password = ...
# escaping special characters such as @ signs
password = urllib.parse.quote_plus(password)
url = f'dialect+driver://username:{password}@host:port/database'
engine = sqlalchemy.create_engine(url)  # Engine object

with engine.connect() as conn:
    results = conn.execute('SELECT * FROM students')
    for row in results:
        print(row)
Read more »

| ~ |

并发 (concurrency) 和并行 (parallelism) 是两个概念. 这两个术语的使用还有争议, 下面依照书上 (Fluent Python, 2nd Edition) 的说法, 并行是并发的子集.

A modern laptop with 4 CPU cores is routinely running more than 200 processes at any given time under normal, casual use. To execute 200 tasks in parallel, you’d need 200 cores. So, in practice, most computing is concurrent and not parallel.

Read more »

| ~ |

首先是这篇文章

内容不再赘述, 下面是纲要和一些对原文的注解. 虽然文章用 Torch 和 GPU 举例, 但基本原理依然普适.

注 (Show more »)

机器之心 编译 过这篇文章, 但有些错误. 比如只看第一段, 前几句话还行, 但光最后一句就有两个技术错误: (1) 把 in-place 说成内置; (2) 梯度设置为 None 说成 0. 关于第 2 点可以参考 torch 文档 Use parameter.grad = None instead of model.zero_grad() or optimizer.zero_grad().

简单介绍了模型耗时可以分为三部分

  • Compute: Time spent on your GPU computing actual floating point operations (FLOPS)
  • Bandwith: Time spent on moving the data from CPU to GPU, from one node to another, or even from CUDA global memory to CUDA shared memory
  • Overhead: Everything else
Read more »

| ~ |

主要参考

学习目标: 能写简单脚本, 能看懂长一些的脚本. Shell 脚本不是开发语言, 难以 debug, 不适合写太长.

If you are writing a script that is more than 100 lines long, or that uses non-straightforward control flow logic, you should rewrite it in a more structured language now. Bear in mind that scripts grow. Rewrite your script early to avoid a more time-consuming rewrite at a later date. (Google Shell Style Guide)

Read more »