xzuyn

Results 21 issues of xzuyn

Specifically in Totally Accurate Battlegrounds (I have not tried with other games as I only need this program for that). When the cross-hair is on screen my game will feel...

bug
wontfix
blocked
Blocked by Electron

added Galaxy S8 Model SM-G950W with sensor width of 5.66

sensordb

My main drive has very little space left, and it seems to default to downloading there. Is it possible to have it install to a separate drive? I'm using Edge...

**Is your feature request related to a problem? Please describe.** ![lllama](https://user-images.githubusercontent.com/16216325/235540124-eab3d410-d62f-4da7-8ab2-45b205cdf436.png) Not exactly a problem, but depending on the model, it naturally likes speech to be in quotes, and the...

[Cerebras Blog Post](https://www.cerebras.net/blog/slimpajama-a-627b-token-cleaned-and-deduplicated-version-of-redpajama) [Hugging Face Dataset](https://huggingface.co/datasets/cerebras/SlimPajama-627B) It's RedPajama, but deduplicated down to 672B tokens using MinHashLSH. May be worth using instead as it should be more compute efficient. ![SlimPajama](https://github.com/openlm-research/open_llama/assets/16216325/05fe4f82-44f5-4263-ba87-6b6dc836f250)

> [2023.10 - We recently release new v4.7-4.8 optimized for anime scenes! 🎉](https://github.com/hzwer/Practical-RIFE) Or at least a little guide on how users can convert future models to NCNN themselves?

enhancement

Instead of having it generate randomly, I would like to have it use a specific phrase (or maybe a list of phrases chosen randomly) to start them.

So I wanna be able to stop training at say 500 steps, then later on continue that from 500 steps into the training data on my saved model. How can...

While some of these instructions are great, and use some very interesting phrasing which I definitely feel could help with training, I don't think I could recommend using it without...

The base GPT-2 is not much to work with in terms of quality. edit: It seems to handle medium, but large runs out of memory. Can some of it possibly...