GitHub - openai/openai-python: The official Python library for the OpenAI API · GitHub
GitHub - openai/openai-python: The official Python library for the OpenAI API · GitHub
29
公开标注数
7
参与人数
2026-07-16 09:47:54
首次 Whisper
讨论活跃度
github.com 近 17 周的公开 Whisper
少多
最新公开 Whisper
划选高亮2026-08-02 09:09:18
原文高亮摘录
“curl http://localhost:11434/api/chat”
Whisper 随想笔记
curl command straight into terminal, love how simple it is.
划选高亮2026-08-01 03:09:18
原文高亮摘录
“standard interface for models, embeddings, vector stores, and more.”
Whisper 随想笔记
Thats the core value prop right there, a standard interface for everything LLM.
划选高亮2026-07-30 21:09:18
原文高亮摘录
“not a modular toolbox of building blocks for neural nets.”
Whisper 随想笔记
That's a refreshing take, keeping it simple for researchers to hack on.
划选高亮2026-07-25 13:07:23
原文高亮摘录
“Make your ML code future-proof by avoiding framework lock-in.”
Whisper 随想笔记
Honestly, just pick one and stick with it, rewrites cost way more time.
划选高亮2026-07-25 12:58:23
原文高亮摘录
“Make your ML code future-proof by avoiding framework lock-in.”
Whisper 随想笔记
Yeah but every framework has its own quirks, porting is never that clean.
划选高亮2026-07-25 10:04:23
原文高亮摘录
“Keras 3 is a multi-backend deep learning framework, with support for JAX, TensorFlow, PyTorch, and OpenVINO”
Whisper 随想笔记
OpenVINO support is nice, but who really uses that for training?
划选高亮2026-07-25 09:55:23
原文高亮摘录
“Keras 3 is a multi-backend deep learning framework, with support for JAX, TensorFlow, PyTorch, and OpenVINO”
Whisper 随想笔记
I've been using Keras with TF for years, this is a game changer.
划选高亮2026-07-25 09:46:23
原文高亮摘录
“Keras 3 is a multi-backend deep learning framework, with support for JAX, TensorFlow, PyTorch, and OpenVINO”
Whisper 随想笔记
Multi-backend sounds great but I worry about lock-in to the Keras API.
划选高亮2026-07-24 13:07:18
原文高亮摘录
“No token limit for prompts (original stable diffusion lets you use up to 75 tokens)”
Whisper 随想笔记
Haha, now I can write my entire fanfic in one prompt.
划选高亮2026-07-24 12:58:18
原文高亮摘录
“No token limit for prompts (original stable diffusion lets you use up to 75 tokens)”
Whisper 随想笔记
Wait, really? I always hit the 75 cap, that's gonna change everything.
划选高亮2026-07-24 10:04:18
原文高亮摘录
“Attention, specify parts of text that the model should pay more attention to”
Whisper 随想笔记
Does the Ctrl+Up thing actually work on Mac? I've been doing it manually all this time.
划选高亮2026-07-24 09:55:18
原文高亮摘录
“Attention, specify parts of text that the model should pay more attention to”
Whisper 随想笔记
I always wondered what those double parentheses did, now I finally get it.
划选高亮2026-07-24 09:46:18
原文高亮摘录
“Attention, specify parts of text that the model should pay more attention to”
Whisper 随想笔记
So that's how you make the model focus on the bowtie instead of the whole suit, neat.
划选高亮2026-07-23 12:39:46
原文高亮摘录
“Seamless integration with popular Hugging Face models”
Whisper 随想笔记
That's the main reason I switched from other engines honestly.
划选高亮2026-07-23 12:30:46
原文高亮摘录
“Seamless integration with popular Hugging Face models”
Whisper 随想笔记
Yeah, but sometimes the custom models still break, you know?
划选高亮2026-07-23 09:36:46
原文高亮摘录
“Efficient management of attention key and value memory with PagedAttention”
Whisper 随想笔记
I remember struggling with OOM errors before this—glad it's open source.
划选高亮2026-07-23 09:27:46
原文高亮摘录
“Efficient management of attention key and value memory with PagedAttention”
Whisper 随想笔记
But does it actually work well with existing models out of the box?
划选高亮2026-07-23 09:18:46
原文高亮摘录
“Efficient management of attention key and value memory with PagedAttention”
Whisper 随想笔记
PagedAttention sounds like a game changer for long context windows.
划选高亮2026-07-22 16:10:00
原文高亮摘录
“These models are not finetuned for chat or Q&A.”
Whisper 随想笔记
So basically they're just raw text generators, not chatbots.
划选高亮2026-07-22 13:07:00
原文高亮摘录
“the links expire after 24 hours and a certain amount of downloads.”
Whisper 随想笔记
Ha, classic—just when you think you got it, the link dies on you.