legraphista's picture
Upload README.md with huggingface_hub
f01b285 verified
metadata
base_model: internlm/internlm2_5-20b-chat
inference: false
library_name: gguf
license: other
pipeline_tag: text-generation
quantized_by: legraphista
tags:
  - quantized
  - GGUF
  - quantization
  - imat
  - imatrix
  - static
  - 16bit
  - 8bit
  - 6bit
  - 5bit
  - 4bit
  - 3bit
  - 2bit
  - 1bit

internlm2_5-20b-chat-IMat-GGUF

Llama.cpp imatrix quantization of internlm/internlm2_5-20b-chat

Original Model: internlm/internlm2_5-20b-chat
Original dtype: BF16 (bfloat16)
Quantized by: llama.cpp b3524
IMatrix dataset: here


Files

IMatrix

Status: βœ… Available
Link: here

Common Quants

Filename Quant type File Size Status Uses IMatrix Is Split
internlm2_5-20b-chat.Q8_0.gguf Q8_0 21.11GB βœ… Available βšͺ Static πŸ“¦ No
internlm2_5-20b-chat.Q6_K.gguf Q6_K 16.30GB βœ… Available βšͺ Static πŸ“¦ No
internlm2_5-20b-chat.Q4_K.gguf Q4_K 11.98GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.Q3_K.gguf Q3_K 9.72GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.Q2_K.gguf Q2_K 7.55GB βœ… Available 🟒 IMatrix πŸ“¦ No

All Quants

Filename Quant type File Size Status Uses IMatrix Is Split
internlm2_5-20b-chat.BF16.gguf BF16 39.73GB βœ… Available βšͺ Static πŸ“¦ No
internlm2_5-20b-chat.FP16.gguf F16 39.73GB βœ… Available βšͺ Static πŸ“¦ No
internlm2_5-20b-chat.Q8_0.gguf Q8_0 21.11GB βœ… Available βšͺ Static πŸ“¦ No
internlm2_5-20b-chat.Q6_K.gguf Q6_K 16.30GB βœ… Available βšͺ Static πŸ“¦ No
internlm2_5-20b-chat.Q5_K.gguf Q5_K 14.08GB βœ… Available βšͺ Static πŸ“¦ No
internlm2_5-20b-chat.Q5_K_S.gguf Q5_K_S 13.73GB βœ… Available βšͺ Static πŸ“¦ No
internlm2_5-20b-chat.Q4_K.gguf Q4_K 11.98GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.Q4_K_S.gguf Q4_K_S 11.40GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.IQ4_NL.gguf IQ4_NL 11.36GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.IQ4_XS.gguf IQ4_XS 10.77GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.Q3_K.gguf Q3_K 9.72GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.Q3_K_L.gguf Q3_K_L 10.55GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.Q3_K_S.gguf Q3_K_S 8.76GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.IQ3_M.gguf IQ3_M 9.12GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.IQ3_S.gguf IQ3_S 8.80GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.IQ3_XS.gguf IQ3_XS 8.36GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.IQ3_XXS.gguf IQ3_XXS 7.81GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.Q2_K.gguf Q2_K 7.55GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.Q2_K_S.gguf Q2_K_S 7.01GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.IQ2_M.gguf IQ2_M 6.97GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.IQ2_S.gguf IQ2_S 6.47GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.IQ2_XS.gguf IQ2_XS 6.10GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.IQ2_XXS.gguf IQ2_XXS 5.54GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.IQ1_M.gguf IQ1_M 4.92GB βœ… Available 🟒 IMatrix πŸ“¦ No
internlm2_5-20b-chat.IQ1_S.gguf IQ1_S 4.54GB 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/internlm2_5-20b-chat-IMat-GGUF --include "internlm2_5-20b-chat.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/internlm2_5-20b-chat-IMat-GGUF --include "internlm2_5-20b-chat.Q8_0/*" --local-dir ./
# see FAQ for merging GGUF's

Inference

Simple chat template

<s><|im_start|>user
{user_prompt}<|im_end|>
<|im_start|>assistant
{assistant_response}<|im_end|>
<|im_start|>user
{next_user_prompt}<|im_end|>

Chat template with system prompt

<s><|im_start|>system
{system_prompt}<|im_end|>
<|im_start|>user
{user_prompt}<|im_end|>
<|im_start|>assistant
{assistant_response}<|im_end|>
<|im_start|>user
{next_user_prompt}<|im_end|>

Llama.cpp

llama.cpp/main -m internlm2_5-20b-chat.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: internlm2_5-20b-chat.Q8_0)
  3. Run gguf-split --merge internlm2_5-20b-chat.Q8_0/internlm2_5-20b-chat.Q8_0-00001-of-XXXXX.gguf internlm2_5-20b-chat.Q8_0.gguf
    • Make sure to point gguf-split to the first chunk of the split.

Got a suggestion? Ping me @legraphista!