Erik Scholz

> Weit gefehlt: Es ist der gleiche Offset. Insgesamt wird auf diesen Block 1400 mal verwiesen, Es handelt sich hier um eine einfache Form von "Kompression", in dem doppelte bilder...

> and that mmap works over a network share. had issues with this in the past (couble of weeks). It first works, I can read the full file no problem,...

> NFS or CIFS? CIFS , between 2 ubuntu machines

A simple finetune (lora is enough) for a stretched rope would be enough. see eg https://github.com/ggerganov/llama.cpp/discussions/1965

sounds like you are not using rope scaling. some rope scaling variants can get away without finetuning.

someone started to train a 1.1B llama, with gqa https://github.com/jzhang38/TinyLlama

https://huggingface.co/chavinlo/Alpaca-65B/tree/main

@d33tah The finetune does not contain the full model. to cite [LoRA](https://github.com/microsoft/LoRA), the used technique: _LoRA reduces the number of trainable parameters by learning pairs of rank-decompostion matrices while freezing...

> it hangs in WaitForThread inside of SDL_Camera Thread inside a call to v4l2, which gdb does not name. I ran into the same issue when I was reading the...

I suspect the YUY2 frame is not fully created, because I sometimes (every ~100th) hit a **heap-buffer-overlow** when blitting from the aquired frame: ``` ### found cameras: - Camera #0:...