Skip to content

Preemo-Inc/onefiveeight

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

20 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

All things 1.58 Bit

Roadmap

  1. Check if we can pretrain from scratchwith 1.58 Bit (random initialized) (we are here)
  2. Initialize 1.58 Bit from Mixtral/Mistral Weights (we are here)
  3. Continued pretraining
  4. Move to ASIC
  5. AGI (in 1.58 bit, on ASIC)

Setup

python3 -m venv venv
. ./venv/bin/activate
cd hqq && pip install -e .

TODO for 1Bit

HQQ -> 1.58

  • test HQQ -> fork -> 1.58bit
  • Compare performance of trained bitnet model with 1.58-hqq-quantized pretrained model
    • Given data $D$: Compare $Bitnet(D)$ with $HQQ_{1.58}(Model_{fp16/fp32}(D))$
  • 2bit quant llama / bitsandbytes

Eval Results

Model Dataset Quant Groupsize PPL
TheBloke/Llama-2-7B-fp16 wikitext + wikitext_wikitext-2-raw-v1, validation splits HQQ 1.58 16 400.46
TheBloke/Llama-2-7B-fp16 wikitext + wikitext_wikitext-2-raw-v1, validation splits HQQ 1.58 8 8.69
TheBloke/Llama-2-7B-fp16 wikitext + wikitext_wikitext-2-raw-v1, validation splits FP16 - 5.18
TheBloke/Llama-2-13B-fp16 wikitext + wikitext_wikitext-2-raw-v1, validation splits HQQ 1.58 16 48.23
TheBloke/Llama-2-13B-fp16 wikitext + wikitext_wikitext-2-raw-v1, validation splits HQQ 1.58 8 7.2732