legraphista's picture
Upload README.md with huggingface_hub
799805a verified
metadata
base_model: failspy/Meta-Llama-3-70B-Instruct-abliterated-v3.5
inference: false
library_name: gguf
license: llama3
pipeline_tag: text-generation
quantized_by: legraphista
tags:
  - quantized
  - GGUF
  - imatrix
  - quantization
  - imat
  - imatrix
  - static
  - 16bit
  - 8bit
  - 6bit
  - 5bit
  - 4bit
  - 3bit
  - 2bit
  - 1bit

Meta-Llama-3-70B-Instruct-abliterated-v3.5-IMat-GGUF

Llama.cpp imatrix quantization of failspy/Meta-Llama-3-70B-Instruct-abliterated-v3.5

Original Model: failspy/Meta-Llama-3-70B-Instruct-abliterated-v3.5
Original dtype: BF16 (bfloat16)
Quantized by: llama.cpp b3058
IMatrix dataset: here


Files

IMatrix

Status: βœ… Available
Link: here

Common Quants

Filename Quant type File Size Status Uses IMatrix Is Split
Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q8_0/* Q8_0 74.98GB βœ… Available βšͺ Static βœ‚ Yes
Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q6_K/* Q6_K 57.89GB βœ… Available βšͺ Static βœ‚ Yes
Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q4_K.gguf Q4_K 42.52GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q3_K.gguf Q3_K 34.27GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q2_K.gguf Q2_K 26.38GB βœ… Available 🟒 IMatrix πŸ“¦ No

All Quants

Filename Quant type File Size Status Uses IMatrix Is Split
Meta-Llama-3-70B-Instruct-abliterated-v3.5.BF16/* BF16 141.12GB βœ… Available βšͺ Static βœ‚ Yes
Meta-Llama-3-70B-Instruct-abliterated-v3.5.FP16/* F16 141.12GB βœ… Available βšͺ Static βœ‚ Yes
Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q8_0/* Q8_0 74.98GB βœ… Available βšͺ Static βœ‚ Yes
Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q6_K/* Q6_K 57.89GB βœ… Available βšͺ Static βœ‚ Yes
Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q5_K/* Q5_K 49.95GB βœ… Available βšͺ Static βœ‚ Yes
Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q5_K_S/* Q5_K_S 48.66GB βœ… Available βšͺ Static βœ‚ Yes
Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q4_K.gguf Q4_K 42.52GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q4_K_S.gguf Q4_K_S 40.35GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.IQ4_NL.gguf IQ4_NL 40.05GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.IQ4_XS.gguf IQ4_XS 37.90GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q3_K.gguf Q3_K 34.27GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q3_K_L.gguf Q3_K_L 37.14GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q3_K_S.gguf Q3_K_S 30.91GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.IQ3_M.gguf IQ3_M 31.94GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.IQ3_S.gguf IQ3_S 30.91GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.IQ3_XS.gguf IQ3_XS 29.31GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.IQ3_XXS.gguf IQ3_XXS 27.47GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q2_K.gguf Q2_K 26.38GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q2_K_S.gguf Q2_K_S 24.47GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.IQ2_M.gguf IQ2_M 24.12GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.IQ2_S.gguf IQ2_S 22.24GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.IQ2_XS.gguf IQ2_XS 21.14GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.IQ2_XXS.gguf IQ2_XXS 19.10GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.IQ1_M.gguf IQ1_M 16.75GB βœ… Available 🟒 IMatrix πŸ“¦ No
Meta-Llama-3-70B-Instruct-abliterated-v3.5.IQ1_S.gguf IQ1_S 15.34GB βœ… Available 🟒 IMatrix πŸ“¦ No

Downloading using huggingface-cli

If you do not have hugginface-cli installed:

pip install -U "huggingface_hub[cli]"

Download the specific file you want:

huggingface-cli download legraphista/Meta-Llama-3-70B-Instruct-abliterated-v3.5-IMat-GGUF --include "Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q8_0.gguf" --local-dir ./

If the model file is big, it has been split into multiple files. In order to download them all to a local folder, run:

huggingface-cli download legraphista/Meta-Llama-3-70B-Instruct-abliterated-v3.5-IMat-GGUF --include "Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q8_0/*" --local-dir ./
# see FAQ for merging GGUF's

Inference

Simple chat template

<|begin_of_text|><|start_header_id|>user<|end_header_id|>

{user_prompt}<|eot_id|><|start_header_id|>assistant<|end_header_id|>

{assistant_response}<|eot_id|><|start_header_id|>user<|end_header_id|>

{next_user_prompt}<|eot_id|>

Chat template with system prompt

<|begin_of_text|><|start_header_id|>system<|end_header_id|>

{system_prompt}<|eot_id|><|start_header_id|>user<|end_header_id|>

{user_prompt}<|eot_id|><|start_header_id|>assistant<|end_header_id|>

{assistant_response}<|eot_id|><|start_header_id|>user<|end_header_id|>

{next_user_prompt}<|eot_id|>

Llama.cpp

llama.cpp/main -m Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q8_0.gguf --color -i -p "prompt here (according to the chat template)"

FAQ

Why is the IMatrix not applied everywhere?

According to this investigation, it appears that lower quantizations are the only ones that benefit from the imatrix input (as per hellaswag results).

How do I merge a split GGUF?

  1. Make sure you have gguf-split available
  2. Locate your GGUF chunks folder (ex: Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q8_0)
  3. Run gguf-split --merge Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q8_0/Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q8_0-00001-of-XXXXX.gguf Meta-Llama-3-70B-Instruct-abliterated-v3.5.Q8_0.gguf
    • Make sure to point gguf-split to the first chunk of the split.

Got a suggestion? Ping me @legraphista!