So Good They Can't Ignore You: Why Skills Trump Passion in the Quest for Work You Love 读书笔记

本书强调了将热情转化为职业生涯中不可或缺的技能的重要性,指出只有掌握人们愿意为之付费的宝贵技能,才能建立稳固的职业基础。书中提到,梦想的工作不会从天而降,需要通过努力工作来创造。成功往往来源于长期的刻意练习,即使过程艰辛,但最终会收获显著成果。作者建议采用工匠心态,专注于提升专业技能,而非过分关注工作是否完美。

Lessons learned from this book:

1. Passion is a vague thing and can lead you astray if your passion is not backed up by your career capital.

2. Master valuable skills that people are willing to pay you for. This is the foundation from which a career and mission can be built on.

3. There is no dream job waiting for us to discover. We have to work hard to create those jobs that we wish to engage in.

书摘:

Glass emphasizes that it takes time to get good at anything, recounting the many years it took him to master radio to the point where he had interesting options. “The key thing is to force yourself through the work, force the skills to come; that’s the hardest phase,” he says.

Glass continues: “I feel like your problem is that you’re trying to judge all things in the abstract before you do them. That’s your tragic mistake.”

It’s hard to predict in advance what you’ll eventually grow to love.

“You’ll never be sure. You don’t want to be sure.”

Compelling careers often have complex origins that reject the simple idea that all you have to do is follow your passion.

As I concluded after meeting Jordan Tice, there’s something liberating about the craftsman mindset: It asks you to leave behind self-centered concerns about whether your job is “just right,” and instead put your head down and plug away at getting really damn good. No one owes you a great career, it argues; you need to earn it—and the process won’t be easy.

“The key thing is to force yourself through the work, force the skills to come; that’s the hardest phase,”

You need to get good in order to get good things in your working life

Deliberate practice is often the opposite of enjoyable.

You stretch yourself, day after day, month after month, before finally looking up and realizing, “Hey, I’ve become pretty good, and people are starting to notice.”

Even with the craftsman mindset, however, becoming “so good they can’t ignore you” is not trivial. To help these efforts I introduced the well-studied concept of deliberate practice, an approach to work where you deliberately stretch your abilities beyond where you’re comfortable and then receive ruthless feedback on your performance. Musicians, athletes, and chess players know all about deliberate practice. Knowledge workers, however, do not. This is great news for knowledge workers: If you can introduce this strategy into your working life you can vault past your peers in your acquisition of career capital.

You have to get good before you can expect good work.

In most jobs you should expect your employer to resist your move toward more control; they have every incentive to try to convince you to reinvest your career capital back into your career at their company, obtaining more money and prestige instead of more control, and this can be a hard argument to resist.

Do what people are willing to pay for. “Money is a neutral indicator of value. By aiming to make money, you’re aiming to be valuable.” --Derek Sivers

随着信息技术在管理上越来越深入而广泛的应用,作为学校以及一些培训机构,都在用信息化战术来部署线上学习以及线上考试,可以与线下的考试有机的结合在一起,实现基于SSM的小码创客教育教学资源库的设计与实现在技术上已成熟。本文介绍了基于SSM的小码创客教育教学资源库的设计与实现的开发全过程。通过分析企业对于基于SSM的小码创客教育教学资源库的设计与实现的需求,创建了一个计算机管理基于SSM的小码创客教育教学资源库的设计与实现的方案。文章介绍了基于SSM的小码创客教育教学资源库的设计与实现的系统分析部分,包括可行性分析等,系统设计部分主要介绍了系统功能设计和数据库设计。 本基于SSM的小码创客教育教学资源库的设计与实现有管理员,校长,教师,学员四个角色。管理员可以管理校长,教师,学员等基本信息,校长角色除了校长管理之外,其他管理员可以操作的校长角色都可以操作。教师可以发布论坛,课件,视频,作业,学员可以查看和下载所有发布的信息,还可以上传作业。因而具有一定的实用性。 本站是一个B/S模式系统,采用Java的SSM框架作为开发技术,MYSQL数据库设计开发,充分保证系统的稳定性。系统具有界面清晰、操作简单,功能齐全的特点,使得基于SSM的小码创客教育教学资源库的设计与实现管理工作系统化、规范化。
### 解决 `from_pretrained` 方法中的尺寸不匹配问题 当使用 Hugging Face 的 Transformers 库加载预训练模型时,如果遇到 `"Size mismatch for embed_out.weight"` 错误,这通常是由于目标模型架构与检查点文件中的权重形状不一致引起的[^1]。为了处理这种错误,可以设置参数 `ignore_mismatched_sizes=True` 来忽略这些大小不匹配的情况。 #### 设置 `ignore_mismatched_sizes=True` 通过将 `ignore_mismatched_sizes` 参数设为 `True`,可以在调用 `from_pretrained` 方法时跳过那些维度不匹配的张量加载过程。以下是具体实现方法: ```python from transformers import AutoModelForCausalLM, AutoTokenizer model_name_or_path = "your_model_checkpoint" tokenizer = AutoTokenizer.from_pretrained(model_name_or_path) try: model = AutoModelForCausalLM.from_pretrained( model_name_or_path, ignore_mismatched_sizes=True # 添加此参数以忽略大小不匹配的问题 ) except Exception as e: print(f"Error loading the model: {e}") ``` 上述代码片段展示了如何在加载模型时传递 `ignore_mismatched_sizes=True` 参数来避免因权重形状不同而导致的异常。 #### 处理量化后的模型 对于已经经过量化处理的模型,在加载过程中可能还需要额外考虑量化参数的影响。例如,如果模型被设计用于支持 FSDP(Fully Sharded Data Parallel),则需要确保正确配置设备以及数据类型的转换逻辑[^2]。下面是一个更复杂的例子,展示如何结合量化器调整参数并完成加载操作: ```python import torch from transformers import AutoModelForCausalLM def load_quantized_model(checkpoint_dir, device="cuda"): model = AutoModelForCausalLM.from_pretrained( checkpoint_dir, low_cpu_mem_usage=True, # 减少 CPU 内存占用 torch_dtype=torch.float16, # 使用较低精度的数据类型加速推理 ignore_mismatched_sizes=True # 跳过大小不匹配的部分 ).to(device) return model quantized_model = load_quantized_model("path/to/quantized/checkpoint", device="cuda") ``` 这段代码不仅设置了 `ignore_mismatched_sizes=True`,还引入了其他优化选项如减少内存消耗 (`low_cpu_mem_usage`) 和指定浮点数精度 (`torch_dtype`),从而更好地适配硬件资源需求。 --- ### 注意事项 尽管启用 `ignore_mismatched_sizes=True` 可能会帮助绕开一些常见问题,但它也可能掩盖潜在的设计缺陷或兼容性隐患。因此建议开发者仔细验证最终模型的行为是否符合预期,并对比原始版本确认性能差异。 ---
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值