Original model: https://huggingface.co/Doctor-Shotgun/Nous-Capybara-limarpv3-34B
Using cleaned pippa.parquet as calibration dataset.
5bpw - Can run 10k context in ~23.3GB VRAM with 8bit-cache option.
Original model: https://huggingface.co/Doctor-Shotgun/Nous-Capybara-limarpv3-34B
Using cleaned pippa.parquet as calibration dataset.
5bpw - Can run 10k context in ~23.3GB VRAM with 8bit-cache option.