Ingest, store, & analyze all types of time series data in a fully-managed, purpose-built database. Keep data forever with low-cost storage and superior data compression. Learn more →
Llama.cpp Alternatives
Similar projects and alternatives to llama.cpp
-
text-generation-webui
A gradio web UI for running Large Language Models like GPT-J 6B, OPT, GALACTICA, LLaMA, and Pygmalion.
-
-
InfluxDB
Access the most powerful time series database as a service. Ingest, store, & analyze all types of time series data in a fully-managed, purpose-built database. Keep data forever with low-cost storage and superior data compression.
-
-
-
-
-
stanford_alpaca
Code and documentation to train Stanford's Alpaca models, and generate the data.
-
SonarLint
Clean code begins in your IDE with SonarLint. Up your coding game and discover issues early. SonarLint is a free plugin that helps you find & fix bugs and security issues from the moment you start writing code. Install from your favorite IDE marketplace today.
-
-
Open-Assistant
OpenAssistant is a chat-based assistant that understands tasks, can interact with third-party systems, and retrieve information dynamically to do so.
-
ChatGLM-6B
ChatGLM-6B:开源双语对话语言模型 | An Open Bilingual Dialogue Language Model
-
-
-
-
-
-
-
basaran
Basaran is an open-source alternative to the OpenAI text completion API. It provides a compatible streaming API for your Hugging Face Transformers-based text generation models.
-
-
-
-
SaaSHub
SaaSHub - Software Alternatives and Reviews. SaaSHub helps you find the best software and product alternatives
llama.cpp reviews and mentions
-
Introducing llamacpp-for-kobold, run llama.cpp locally with a fancy web UI, persistent stories, editing tools, save formats, memory, world info, author's note, characters, scenarios and more with minimal setup.
I cannot fix it myself - please raise awareness to it here: https://github.com/ggerganov/llama.cpp/discussions/229
There's an important fix for 65B models upstream: https://github.com/ggerganov/llama.cpp/pull/438/files. I've verified it works on my local copy. Can your fork be updated from upstream? Without it llama will segfault to an under-estimate of the memory required.
-
Cformers 🚀 - "Transformers with a C-backend for lightning-fast CPU inference". | Nolano
How does the inference speed compare to GGML based approaches like llama.cpp? Repo link: https://github.com/ggerganov/llama.cpp
-
GitHub announces a bunch of new GPT-4 powered coding assistants. What should and could Emacs and open-source community do?
There are some "open-source" alternatives for the open-AI generative LLMs there, like LLAMA https://github.com/ggerganov/llama.cpp trained and leaked by Meta, and made available in c++ and actually runnable even on a MacBook Air. Sure it's not as great as Codex, GPT-4 and such, but it's a start (there are also open datasets with lots of code https://huggingface.co/datasets/bigcode/the-stack, so maybe in the future the gap will close).
-
[P] fastLLaMa, A python wrapper to run llama.cpp
llama.cpp as of very recently support `--perplexity`, which calculates the perplexity over the prompt (and you can just pass wikitext2 test set as the prompt for example). See https://github.com/ggerganov/llama.cpp/pull/270
-
Bill Gates just published a 7-page letter about AI and his predictions for its future
https://github.com/ggerganov/llama.cpp [CPU loading with comparatively low memory requirements (7b running on phones and Raspberry Pi 4) - no fancy front end yet]
-
[Project] Alpaca-30B: Facebook's 30b parameter LLaMa fine-tuned on the Alpaca dataset
If my understanding is correct, Alpaca.cpp or https://github.com/ggerganov/llama.cpp are a sort of 'front-end' for these model.
LLaMa/Alpaca work just fine on CPU with llama.cpp/alpaca.cpp. Not very snappy but fast enough for me.
You can run llama-30B on a CPU using llama.cpp, it's just slow. The alpaca models I've seen are the same size as the llama model they are trained on, so I would expect running the alpaca-30B models will be possible on any system capable of running llama-30B.
- LLaMa-cpp: RISC-V (TH1520&D1) benchmark and hack for <1GB DDR device
-
A note from our sponsor - InfluxDB
www.influxdata.com | 25 Mar 2023
Stats
ggerganov/llama.cpp is an open source project licensed under MIT License which is an OSI approved license.