AlexBuz / llama-zip

LLM-powered lossless compression tool

AlexBuz/llama-zip Issues

how to make n_gpu_layers work with cuda?
Closed 8 days ago1
possible to provide benchmark for these?
Closed 8 days ago1
generate binary (for better compression) instead of text
Closed 8 days ago2
ever tried tinyllama or smaller llamas?
Closed 8 days ago1
ollama version?
Closed a month ago2
Does not work on all files because of utf-8 error
Closed a month ago4
Request: Add newest cmix compressor to comparison table.
Closed a month ago1
compare with brotli
Closed a month ago1
Interesting side effect of decompression - original training data extraction
Closed a month ago5
Interactive mode generate text instead of compressing when using specific text
Closed a month ago2
Using the colab in the repository, we can notice it does not use the GPU
Closed a month ago1
Question
Closed a month ago2
Using different models like Phi-3
Closed a month ago7
Post compression ratios
Closed a month ago15
Gibberish produced on 1 word spaces
Closed a month ago2
Perform compression in batches for texts exceeding the 8192 token limit of llama3.
Closed a month ago6