语默 (@ymyumo) 's Twitter Profile
语默

@ymyumo

Love NPL

ID: 910982587

calendar_today28-10-2012 19:42:04

98 Tweet

9 Followers

255 Following

merve (@mervenoyann) 's Twitter Profile Photo

Do you want to generate videos from text? 🍿🎥 A team from JAX Diffusers sprint has trained a model to do so! 🤩💜 Play with it here 👉 huggingface.co/spaces/TempoFu… A better model is in cooking too! 🧑‍🍳 Lopho 👁️👄👁️

Do you want to generate videos from text? 🍿🎥 
A team from JAX Diffusers sprint has trained a model to do so! 🤩💜 
Play with it here 👉  huggingface.co/spaces/TempoFu…
A better model is in cooking too! 🧑‍🍳 <a href="/lopho/">Lopho 👁️👄👁️</a>
ruanyf (@ruanyf) 's Twitter Profile Photo

《WebGPU 为什么重要》,推荐这篇长文。cohost.org/mcc/post/14061… 操作系统的图形 API,目前是不统一的:Windows 是 DirectX,苹果是 Metal,Linux 是 Vulkan。而新发布的 WebGPU 是跨平台的图形编程接口。

《WebGPU 为什么重要》,推荐这篇长文。cohost.org/mcc/post/14061…

操作系统的图形 API,目前是不统一的:Windows 是 DirectX,苹果是 Metal,Linux 是 Vulkan。而新发布的 WebGPU 是跨平台的图形编程接口。
BigCode (@bigcodeproject) 's Twitter Profile Photo

Introducing: 💫StarCoder StarCoder is a 15B LLM for code with 8k context and trained only on permissive data in 80+ programming languages. It can be prompted to reach 40% pass@1 on HumanEval and act as a Tech Assistant. Try it here: shorturl.at/cYZ06r Release thread🧵

Introducing: 💫StarCoder

StarCoder is a 15B LLM for code with 8k context and trained only on permissive data in 80+ programming languages. It can be prompted to reach 40% pass@1 on HumanEval and act as a Tech Assistant.

Try it here: shorturl.at/cYZ06r

Release thread🧵
Sully (@sullyomarr) 's Twitter Profile Photo

Vector databases & embeddings are the current hot thing in AI. Pinecone, a vector DB company, just raised $100M at ~1b valuation. Shopify, Brex, Hubspot and others use them for their AI apps But what are they, how do they work and why are they SO crucial in AI? Let's find out

Vector databases &amp; embeddings are the current hot thing in AI. 

Pinecone, a vector DB company, just raised $100M at ~1b valuation.

Shopify, Brex, Hubspot and others use them for their AI apps

But what are they, how do they work and why are they SO crucial in AI? Let's find out
Cameron R. Wolfe, Ph.D. (@cwolferesearch) 's Twitter Profile Photo

We can use few-shot learning and instruction prompting to solve many problems with large language models (LLMs), but what should we do when these techniques fall short? Here are three advanced prompting approaches that can be used to solve complex problems with LLMs… 🧵 [1/8]

We can use few-shot learning and instruction prompting to solve many problems with large language models (LLMs), but what should we do when these techniques fall short? Here are three advanced prompting approaches that can be used to solve complex problems with LLMs… 🧵 [1/8]
AI Will (@financeyf5) 's Twitter Profile Photo

谷歌发布会内容总结脑图版本! 最大的是Palm2模型, 其次是DuetAI给谷歌办公系赋能(类似微软全家桶copliot),和Bard(类似微软bing搜索),以及很多小功能

谷歌发布会内容总结脑图版本!
最大的是Palm2模型,
其次是DuetAI给谷歌办公系赋能(类似微软全家桶copliot),和Bard(类似微软bing搜索),以及很多小功能
peter! 🥷 (@pwang_szn) 's Twitter Profile Photo

Memory in @LangChainAI is a big deal. It was a lot to go through, but I broke it into bite-size chunks. • 4 Memory Types Explained • How to use Memory in a Chain • How to add Memory to an Agent Save this ↓

Memory in @LangChainAI is a big deal.

It was a lot to go through, but I broke it into bite-size chunks.

• 4 Memory Types Explained
• How to use Memory in a Chain
• How to add Memory to an Agent

Save this ↓
Cameron R. Wolfe, Ph.D. (@cwolferesearch) 's Twitter Profile Photo

Traditionally, LLMs have struggled to solve complex problems that require reasoning. Chain of thought prompting has improved their abilities in this domain, but why stop there? Here are four prompting techniques for solving difficult, multi-step problems with LLMs… 🧵 [1/8]

Traditionally, LLMs have struggled to solve complex problems that require reasoning. Chain of thought prompting has improved their abilities in this domain, but why stop there? Here are four prompting techniques for solving difficult, multi-step problems with LLMs… 🧵 [1/8]
宝玉 (@dotey) 's Twitter Profile Photo

现在ChatGPT的GPT-4应该是32K版本的了,会话长度可以很长了。之前我在学习Google那篇《Enabling conversational interaction on mobile with LLMs》的时候,一开始没看太明白了,于是我把完整内容都喂给了ChatGPT(GPT-4),然后让它帮我总结、翻译,才算搞明白了,比ChatPDF这种向量搜索效果强多了。

现在ChatGPT的GPT-4应该是32K版本的了,会话长度可以很长了。之前我在学习Google那篇《Enabling conversational interaction on mobile with LLMs》的时候,一开始没看太明白了,于是我把完整内容都喂给了ChatGPT(GPT-4),然后让它帮我总结、翻译,才算搞明白了,比ChatPDF这种向量搜索效果强多了。
Harrison Chase (@hwchase17) 's Twitter Profile Photo

🌟privateGPT🌟 - this is sick!!! I've always had people asking me if it was feasible to use LangChain with open source models, and my answer was "at the moment, not really..." but Iván Martínez did it!!!!! this is a huge step forward 👏👏👏 github.com/imartinez/priv…

Hugging Face (@huggingface) 's Twitter Profile Photo

The first RNN in transformers! 🤯 Announcing the integration of RWKV models in transformers with BlinkDL and RWKV community! RWKV is an attention free model that combines the best from RNNs and transformers. Learn more about the model in this blogpost: huggingface.co/blog/rwkv

The first RNN in transformers! 🤯
Announcing the integration of RWKV models in transformers with <a href="/BlinkDL_AI/">BlinkDL</a> and RWKV community!
RWKV is an attention free model that combines the best from RNNs and transformers.
Learn more about the model in this blogpost: huggingface.co/blog/rwkv
Aran Komatsuzaki (@arankomatsuzaki) 's Twitter Profile Photo

Pre-Training to Learn in Context Proposes PICL, a framework to enhance the LMs' in-context learning ability by pre-training on "intrinsic tasks" in the general plain-text corpus using the simple LM objective. repo: github.com/thu-coai/PICL abs: arxiv.org/abs/2305.09137

Pre-Training to Learn in Context

Proposes PICL, a framework to enhance the LMs' in-context learning ability by pre-training on "intrinsic tasks" in the general plain-text corpus using the simple LM objective.

repo: github.com/thu-coai/PICL
abs: arxiv.org/abs/2305.09137
Tw93 (@hitw93) 's Twitter Profile Photo

#工程师学习 很多学习英语的同学,一般会有一层翻译思维,也即先将自己想到的中文在脑袋里先翻译成英文再去思考,而非直接用英语去思考。Rachel's English 这个视频「如何用英语思考|不再需要翻译」很值得学习,提到了 6 个有用的思路我觉得超级有用,需要结合视频来看 🤖 youtube.com/watch?v=SJOnhW…

#工程师学习 很多学习英语的同学,一般会有一层翻译思维,也即先将自己想到的中文在脑袋里先翻译成英文再去思考,而非直接用英语去思考。Rachel's English 这个视频「如何用英语思考|不再需要翻译」很值得学习,提到了 6 个有用的思路我觉得超级有用,需要结合视频来看 
🤖 youtube.com/watch?v=SJOnhW…
Harrison Chase (@hwchase17) 's Twitter Profile Photo

📃New document loader📃 psychic.dev is a platform for integrating with your customer’s SaaS tools like Notion, Zendesk, Confluence, and Google Drive via OAuth Now usable as a DocumentLoader in thanks to @AyanBandyop123! Docs: github.com/hwchase17/lang…

📃New document loader📃

psychic.dev is a platform for integrating with your customer’s SaaS tools like Notion, Zendesk, Confluence, and Google Drive via OAuth

Now usable as a DocumentLoader in <a href="/langchain/"></a> thanks to @AyanBandyop123!

Docs: github.com/hwchase17/lang…
Cos(余弦)😶‍🌫️ (@evilcos) 's Twitter Profile Photo

Crypto 钓鱼目前流行的手法大概有: - eth_sign/personal_sign/eth_signTypedData_* 这种原生签名利用 - Token/NFT 类似 approve/permit 这些函数的利用 - 类似 Uniswap swapExactTokensForTokens/permit2 等这些强大函数的利用 - OpenSea/Blur 等协议函数的利用(五花八门) - TX data 4byte 利用 -