WizardLM VS gpt4all

This is interesting work, and a good contribution, but there is no need to mislead people.

[1] https://github.com/nlpxucan/WizardLM

If you pair this with the latest WizardCoder models, which have a fairly better performance than the standard Salesforce Codegen2 and Codegen2.5, you have a pretty solid alternative to GitHub Copilot that runs completely locally.

(We will update the demo links in our github.)

In terms of speed, we're talking about 140t/s for 7B models, and 40t/s for 33B models on a 3090/4090 now.[1] (1 token ~= 0.75 word) It's quite zippy. llama.cpp performs close on Nvidia GPUs now (but they don't have a handy chart) and you can get decent performance on 13B models on M1/M2 Macs.

You can take a look at a list of evals here: https://llm-tracker.info/books/evals/page/list-of-evals - for general usage, I think home-rolled evals like llm-jeopardy [2] and local-llm-comparison [3] by hobbyists are more useful than most of the benchmark rankings.

That being said, personally I mostly use GPT-4 for code assistance to that's what I'm most interested in, and the latest code assistants are scoring quite well: https://github.com/abacaj/code-eval - a recent replit-3b fine tune the human-eval results for open models (as a point of reference, GPT-3.5 gets 60.4 on pass@1 and 68.9 on pass@10 [4]) - I've only just started playing around with it since replit model tooling is not as good as llamas (doc here: https://llm-tracker.info/books/howto-guides/page/replit-mode...).

I'm interested in potentially applying reflexion or some of the other techniques that have been tried to even further increase coding abilities. (InterCode in particular has caught my eye https://intercode-benchmark.github.io/)

[1] https://github.com/turboderp/exllama#results-so-far

[2] https://github.com/aigoopy/llm-jeopardy

[3] https://github.com/Troyanovsky/Local-LLM-comparison/tree/mai...

[4] https://github.com/nlpxucan/WizardLM/tree/main/WizardCoder

You talking about this? https://github.com/nlpxucan/WizardLM

The smallest model that is close to competent at code is WizardCoder 15B.. https://github.com/nlpxucan/WizardLM/

WizardCoder: Empowering Code Large Language Models with Evol-Instruct (https://github.com/nlpxucan/WizardLM/tree/main/WizardCoder)

I don’t know if Ollama can do this but https://gpt4all.io/ can.

Gpt4all is a local desktop app with a Python API that can be trained on your documents: https://gpt4all.io/

The readme page is cryptic. What does 'mature' mean in this context? What is the sample text a continuation of?

Hving a gif the thing in use would be great, similar to the gpt4all readme page. (https://github.com/nomic-ai/gpt4all)

Check https://github.com/nomic-ai/gpt4all instead.

"I ran performance tests on two systems, here's the results of system 1, and heres the results of system 2. Summarize the results, and build a markdown table containing x,y,z rows."

"extract the reusable functions out of this bash script"

"write me a cfssl command to generate a intermediate CA"

"What is the regex for _____"

"Here are my accomplishments over the last 6 months, summarize them into a 1 page performance report."

etc etc etc

If you're not using GPT4 or some LLM as part of your daily flow you're working too hard.

Get GPT4All (https://gpt4all.io), log into OpenAI, drop $20 on your account, get a API key, and start using GPT4.

offline versija, nors ir ne tokia pažengus - https://github.com/nomic-ai/gpt4all ; https://gpt4all.io/index.html

Take a look at https://gpt4all.io, their docs are pretty awesome

I run one. I built an iMessage-like frontend to it using plain JS and a Python websocket backend. I mostly just use it for curiosity and playing with different prompts. I only have 16GB of RAM to dedicate to it, so I use an 8B parameter model which is enough for fun and chitchat, but I don't find it good enough to replace ChatGPT.

https://github.com/nomic-ai/gpt4all