Explore tweets tagged as #Llama2
Finally wrapped up my llama2.c port to rust last night. I also wrote a long technical blog going through each step! Imo this is the best way to learn anything. https://t.co/NO1McpylHm
0
1
3
llama4が炎上しすぎて元Meta社員がLinkedInに「llama2/3の開発はしましたがllama4は関わってない」と明記する程に。DeepSeek以後上層部が焦り、Metaで開発者にとって環境が悪くなったと聞く。実力を超えるPRは身を滅ぼす。PRしすぎて期待値が上がりすぎてるAI開発会社は日本でも避けたほうがいい。
4
106
484
Just become the world’s first to run an LLM locally on a business card lol. Why? For fun. Details: - llama2.c by @karpathy with modifications from @_davebennett - 24tok/sec on esp32s3 n16r8 w/ a 260k param model - designed the card myself, whole inspiration comes from ouija
123
305
4K
minimalist Rust ML framework with fast demos like whisper and LLaMA2
2
7
65
中国産GPUの衝撃！ 中国は次世代の「オイル」、すなわち計算資源を着実に獲得しつつある 平頭哥PPUは96GB HBM2e、PCIe5.0、400Wという仕様でNVIDIA A800を上回り、Llama2-70B推論では、スループット 2800 tokens/s、H20比+18%。エネルギー効率はH20比で30%以上改善という ソース↓
1
1
6
Most LLM tutorials teach you to use models. Happy-LLM teaches you to build them from scratch, including a complete LLaMA2 implementation with just 215M parameters you can train yourself.
1
0
0
✅Day 46 of my #BuildingChallenge : 🚀 Just built a production-grade LLM API using LangChain + FastAPI! ✨ Key features: - Deploy different LLMs (OpenAI/Llama2) as REST API endpoints. - Secure API key management. - Built a Streamlit client for testing. - Production-ready with
1
1
6
Free LangChain series. Link in comment. Learn how to build end-to-end LLM applications from scratch, covering everything from basic prerequisites to advanced projects like PDF Q&A systems, conversational chatbots, and resume tracking systems using OpenAI, LLAMA2, and Google
2
20
98
been really busy the past week due to moving out-related errands, but think I should be done with the llama2 architecture maybe by today or tomorrow see you soon :)
1
0
49
took me a while because life but finally completed llama2 architecture today will study a bit of prompt processing vs generation phases & then onto llama3 I go
5
3
107
🔥 LLama2-Accessory It is an open-source toolkit for pretraining, finetuning, and deployment of Large Language Models (LLMs) and multimodal LLMs. This repo is mainly inherited from LLaMA-Adapter with more advanced features. Repo Link - https://t.co/lAfwQz6RPR
0
0
0
Meta’s Chief AI Scientist Yann LeCun says the quiet part out loud: Without Meta’s AI, there would have been no DeepSeek 🇨🇳 “Without Llama, no AI sovereignty outside the US. Without Llama2 and Llama3, no DeepSeek.” “Without Llama (or PyTorch) there would have been no DeepSeek.”
@gpjt @ClementDelangue Without Llama (or PyTorch) there would have been no DeepSeek. The point of open source is that you can accelerate progress by building on each other's work.
10
5
80
All deepseek model also use the same optimizer parameter for adamW - beta1= 0.9 - beta2= 0.95 - weight decay= 0.1 - gradient clipping = 1.0 Similar to llama2 tech report value
2
2
19
end of any possible US DeepSeek/Kimi moment? Not sure who else has the appetite and scale Llama2 was the first ~os model series i was genuinely surprised & impressed by, and Llama3 was a banger. Honestly felt the Meta LLMs could be the standard
0
0
2
Day 26 of Learning LLMs And Finally today I release the first model of GujjuGPT About This model - It is the first instruct model of gujjuGPT(v1) - Base model used llama2-7b - It is a gujarati based language model Hugging face repo:
1
1
7
TinyScaleLab update: Completed initial training runs to estimate costs. Using 4 model sizes: 5M, 25M, 60M, and 125M parameters. Training on TinyStories dataset (4.9M stories, ~1B tokens) with Llama2 tokenizer (32k vocab).
1
0
1
Why @ritualnet is the Next Big Thing? Ritual is the first sovereign AI execution layer that makes building smart, decentralized apps easy, fast, and secure powered by AI models like LLAMA2 and Mistral.
12
0
48