EasierAI/Granite-3.2-2B
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

初始化项目，由ModelHub XC社区提供模型

Browse Source 
Model: EasierAI/Granite-3.2-2B
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-10-13 08:26:17 +08:00
commit 25bd1e0892
18 changed files with 296 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

75
.gitattributes vendored Normal file
View File

@@ -0,0 +1,75 @@
*.7z filter=lfs diff=lfs merge=lfs -text
*.arrow filter=lfs diff=lfs merge=lfs -text
*.bin filter=lfs diff=lfs merge=lfs -text
*.bz2 filter=lfs diff=lfs merge=lfs -text
*.ckpt filter=lfs diff=lfs merge=lfs -text
*.ftz filter=lfs diff=lfs merge=lfs -text
*.gz filter=lfs diff=lfs merge=lfs -text
*.h5 filter=lfs diff=lfs merge=lfs -text
*.joblib filter=lfs diff=lfs merge=lfs -text
*.lfs.* filter=lfs diff=lfs merge=lfs -text
*.mlmodel filter=lfs diff=lfs merge=lfs -text
*.model filter=lfs diff=lfs merge=lfs -text
*.msgpack filter=lfs diff=lfs merge=lfs -text
*.npy filter=lfs diff=lfs merge=lfs -text
*.npz filter=lfs diff=lfs merge=lfs -text
*.onnx filter=lfs diff=lfs merge=lfs -text
*.ot filter=lfs diff=lfs merge=lfs -text
*.parquet filter=lfs diff=lfs merge=lfs -text
*.pb filter=lfs diff=lfs merge=lfs -text
*.pickle filter=lfs diff=lfs merge=lfs -text
*.pkl filter=lfs diff=lfs merge=lfs -text
*.pt filter=lfs diff=lfs merge=lfs -text
*.pth filter=lfs diff=lfs merge=lfs -text
*.rar filter=lfs diff=lfs merge=lfs -text
*.safetensors filter=lfs diff=lfs merge=lfs -text
saved_model/**/* filter=lfs diff=lfs merge=lfs -text
*.tar.* filter=lfs diff=lfs merge=lfs -text
*.tar filter=lfs diff=lfs merge=lfs -text
*.tflite filter=lfs diff=lfs merge=lfs -text
*.tgz filter=lfs diff=lfs merge=lfs -text
*.wasm filter=lfs diff=lfs merge=lfs -text
*.xz filter=lfs diff=lfs merge=lfs -text
*.zip filter=lfs diff=lfs merge=lfs -text
*.zst filter=lfs diff=lfs merge=lfs -text
*tfevents* filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q8_0.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q6_K_L.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q6_K.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q5_K_L.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q5_K_M.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q5_K_S.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q4_K_L.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q4_K_M.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q4_K_S.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q4_1.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q4_0.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-IQ4_NL.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-IQ4_XS.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q3_K_XL.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q3_K_L.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q3_K_M.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-IQ3_M.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q3_K_S.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-IQ3_XS.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-IQ3_XXS.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q2_K_L.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-Q2_K.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct-IQ2_M.gguf filter=lfs diff=lfs merge=lfs -text
ibm-granite_granite-3.2-2b-instruct.imatrix filter=lfs diff=lfs merge=lfs -text
Granite-3.2-2B-Q2_K.gguf filter=lfs diff=lfs merge=lfs -text
Granite-3.2-2B-Q2_K_L.gguf filter=lfs diff=lfs merge=lfs -text
Granite-3.2-2B-Q3_K_L.gguf filter=lfs diff=lfs merge=lfs -text
Granite-3.2-2B-Q3_K_M.gguf filter=lfs diff=lfs merge=lfs -text
Granite-3.2-2B-Q3_K_S.gguf filter=lfs diff=lfs merge=lfs -text
Granite-3.2-2B-Q3_K_XL.gguf filter=lfs diff=lfs merge=lfs -text
Granite-3.2-2B-Q4_0.gguf filter=lfs diff=lfs merge=lfs -text
Granite-3.2-2B-Q4_K_L.gguf filter=lfs diff=lfs merge=lfs -text
Granite-3.2-2B-Q4_K_M.gguf filter=lfs diff=lfs merge=lfs -text
Granite-3.2-2B-Q4_K_S.gguf filter=lfs diff=lfs merge=lfs -text
Granite-3.2-2B-Q5_K_L.gguf filter=lfs diff=lfs merge=lfs -text
Granite-3.2-2B-Q5_K_M.gguf filter=lfs diff=lfs merge=lfs -text
Granite-3.2-2B-Q5_K_S.gguf filter=lfs diff=lfs merge=lfs -text
Granite-3.2-2B-Q6_K.gguf filter=lfs diff=lfs merge=lfs -text
Granite-3.2-2B-Q6_K_L.gguf filter=lfs diff=lfs merge=lfs -text
Granite-3.2-2B-Q8_0.gguf filter=lfs diff=lfs merge=lfs -text

3
Granite-3.2-2B-Q2_K.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:949d1057c9406afc65d80418cacbf15474378bac899ff42327ac6056dc6ebb06
size 978246272

3
Granite-3.2-2B-Q2_K_L.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:7b8eff31dd0608494c464120ea27043ae72b5fcb4028ef887ed7b207133032c4
size 1002627136

3
Granite-3.2-2B-Q3_K_L.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:e63a19be192cd498e4bad96f89a42d1545cfb1ca4b24fb96702b96f628886294
size 1357372032

3
Granite-3.2-2B-Q3_K_M.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:82fe988308c1573a336b7f2b4ce0c406f859124b7f563aba2c8b1e9b177e4d6d
size 1251728000

3
Granite-3.2-2B-Q3_K_S.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:255b9ee595f3a0f1b8ac4785e876002aa367b80d6d01cd5830670bfc30b6e4bb
size 1130289792

3
Granite-3.2-2B-Q4_0.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:8501fcc69b0b4ff98d96ca24861a0acdcf2a7e42aebd1bdea6be6ba1700a6098
size 1458625152

3
Granite-3.2-2B-Q4_K_L.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:9db4205ebac892b3c97a1c54baa51e7d5a0352b6c5fe380245272f37e76f8b7b
size 1569677376

3
Granite-3.2-2B-Q4_K_M.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:e1b915b0849becf4fdda188dee7b09cbebbfabd71c6f3f2b75dd3eca0a8fded1
size 1545296512

3
Granite-3.2-2B-Q4_K_S.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:c272fb73fcf5cf5dc96be63e361339d8eed4333cf1bfcca01fc055939d8d2694
size 1464392320

3
Granite-3.2-2B-Q5_K_L.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:f981c919b3d92a1f75971044afa1e0f216f7b041a8e097d27778c7434b4d9ad4
size 1829199936

3
Granite-3.2-2B-Q5_K_M.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:40b74fae1d3d0ac2872e5dde6d8d88423c653fff20b30fdd1f00db0394a2f34f
size 1804819072

3
Granite-3.2-2B-Q5_K_S.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:5ea9d1f52ac2255c285f25f5d550e24c0c5ff818641a67370c4436532acc0d3f
size 1757469312

3
Granite-3.2-2B-Q6_K.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:c87464fde6e9c4388aeaaa01b9e730839cb3d55a3c3e7d9c615409cd5d4a0824
size 2080561792

3
Granite-3.2-2B-Q6_K_L.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:36a7c2c068732e037b34c5cc3592cb5d193bb9ff54c01284e2fc5989a1449e3a
size 2104942656

3
Granite-3.2-2B-Q8_0.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:7ffbd0fe17ac37775c3758464aa9a09773a3a162b9459eb9094278a7a809682a
size 2694111296

173
README.md Normal file
View File

@@ -0,0 +1,173 @@
---
quantized_by: bartowski
pipeline_tag: text-generation
base_model: ibm-granite/granite-3.2-2b-instruct
license: apache-2.0
tags:
- language
- granite-3.2
inference: false
---
## Llamacpp imatrix Quantizations of granite-3.2-2b-instruct by ibm-granite
Using <a href="https://github.com/ggerganov/llama.cpp/">llama.cpp</a> release <a href="https://github.com/ggerganov/llama.cpp/releases/tag/b4778">b4778</a> for quantization.
Original model: https://huggingface.co/ibm-granite/granite-3.2-2b-instruct
All quants made using imatrix option with dataset from [here](https://gist.github.com/bartowski1182/eb213dccb3571f863da82e99418f81e8)
Run them in [LM Studio](https://lmstudio.ai/)
Run them directly with [llama.cpp](https://github.com/ggerganov/llama.cpp), or any other llama.cpp based project
## Prompt format
```
<|start_of_role|>system<|end_of_role|>{system_prompt}<|end_of_text|>
<|start_of_role|>user<|end_of_role|>{prompt}<|end_of_text|>
<|start_of_role|>assistant<|end_of_role|>
```
## Download a file (not the whole branch) from below:
| Filename | Quant type | File Size | Split | Description |
| -------- | ---------- | --------- | ----- | ----------- |
| [granite-3.2-2b-instruct-Q8_0.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q8_0.gguf) | Q8_0 | 2.69GB | false | Extremely high quality, generally unneeded but max available quant. |
| [granite-3.2-2b-instruct-Q6_K_L.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q6_K_L.gguf) | Q6_K_L | 2.10GB | false | Uses Q8_0 for embed and output weights. Very high quality, near perfect, *recommended*. |
| [granite-3.2-2b-instruct-Q6_K.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q6_K.gguf) | Q6_K | 2.08GB | false | Very high quality, near perfect, *recommended*. |
| [granite-3.2-2b-instruct-Q5_K_L.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q5_K_L.gguf) | Q5_K_L | 1.83GB | false | Uses Q8_0 for embed and output weights. High quality, *recommended*. |
| [granite-3.2-2b-instruct-Q5_K_M.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q5_K_M.gguf) | Q5_K_M | 1.80GB | false | High quality, *recommended*. |
| [granite-3.2-2b-instruct-Q5_K_S.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q5_K_S.gguf) | Q5_K_S | 1.76GB | false | High quality, *recommended*. |
| [granite-3.2-2b-instruct-Q4_1.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q4_1.gguf) | Q4_1 | 1.61GB | false | Legacy format, similar performance to Q4_K_S but with improved tokens/watt on Apple silicon. |
| [granite-3.2-2b-instruct-Q4_K_L.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q4_K_L.gguf) | Q4_K_L | 1.57GB | false | Uses Q8_0 for embed and output weights. Good quality, *recommended*. |
| [granite-3.2-2b-instruct-Q4_K_M.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q4_K_M.gguf) | Q4_K_M | 1.55GB | false | Good quality, default size for most use cases, *recommended*. |
| [granite-3.2-2b-instruct-Q4_K_S.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q4_K_S.gguf) | Q4_K_S | 1.46GB | false | Slightly lower quality with more space savings, *recommended*. |
| [granite-3.2-2b-instruct-Q4_0.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q4_0.gguf) | Q4_0 | 1.46GB | false | Legacy format, offers online repacking for ARM and AVX CPU inference. |
| [granite-3.2-2b-instruct-IQ4_NL.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-IQ4_NL.gguf) | IQ4_NL | 1.46GB | false | Similar to IQ4_XS, but slightly larger. Offers online repacking for ARM CPU inference. |
| [granite-3.2-2b-instruct-IQ4_XS.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-IQ4_XS.gguf) | IQ4_XS | 1.38GB | false | Decent quality, smaller than Q4_K_S with similar performance, *recommended*. |
| [granite-3.2-2b-instruct-Q3_K_XL.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q3_K_XL.gguf) | Q3_K_XL | 1.38GB | false | Uses Q8_0 for embed and output weights. Lower quality but usable, good for low RAM availability. |
| [granite-3.2-2b-instruct-Q3_K_L.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q3_K_L.gguf) | Q3_K_L | 1.36GB | false | Lower quality but usable, good for low RAM availability. |
| [granite-3.2-2b-instruct-Q3_K_M.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q3_K_M.gguf) | Q3_K_M | 1.25GB | false | Low quality. |
| [granite-3.2-2b-instruct-IQ3_M.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-IQ3_M.gguf) | IQ3_M | 1.17GB | false | Medium-low quality, new method with decent performance comparable to Q3_K_M. |
| [granite-3.2-2b-instruct-Q3_K_S.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q3_K_S.gguf) | Q3_K_S | 1.13GB | false | Low quality, not recommended. |
| [granite-3.2-2b-instruct-IQ3_XS.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-IQ3_XS.gguf) | IQ3_XS | 1.08GB | false | Lower quality, new method with decent performance, slightly better than Q3_K_S. |
| [granite-3.2-2b-instruct-IQ3_XXS.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-IQ3_XXS.gguf) | IQ3_XXS | 1.01GB | false | Lower quality, new method with decent performance, comparable to Q3 quants. |
| [granite-3.2-2b-instruct-Q2_K_L.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q2_K_L.gguf) | Q2_K_L | 1.00GB | false | Uses Q8_0 for embed and output weights. Very low quality but surprisingly usable. |
| [granite-3.2-2b-instruct-Q2_K.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-Q2_K.gguf) | Q2_K | 0.98GB | false | Very low quality but surprisingly usable. |
| [granite-3.2-2b-instruct-IQ2_M.gguf](https://huggingface.co/bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF/blob/main/ibm-granite_granite-3.2-2b-instruct-IQ2_M.gguf) | IQ2_M | 0.89GB | false | Relatively low quality, uses SOTA techniques to be surprisingly usable. |
## Embed/output weights
Some of these quants (Q3_K_XL, Q4_K_L etc) are the standard quantization method with the embeddings and output weights quantized to Q8_0 instead of what they would normally default to.
## Downloading using huggingface-cli
<details>
<summary>Click to view download instructions</summary>
First, make sure you have hugginface-cli installed:
```
pip install -U "huggingface_hub[cli]"
```
Then, you can target the specific file you want:
```
huggingface-cli download bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF --include "ibm-granite_granite-3.2-2b-instruct-Q4_K_M.gguf" --local-dir ./
```
If the model is bigger than 50GB, it will have been split into multiple files. In order to download them all to a local folder, run:
```
huggingface-cli download bartowski/ibm-granite_granite-3.2-2b-instruct-GGUF --include "ibm-granite_granite-3.2-2b-instruct-Q8_0/*" --local-dir ./
```
You can either specify a new local-dir (ibm-granite_granite-3.2-2b-instruct-Q8_0) or download them all in place (./)
</details>
## ARM/AVX information
Previously, you would download Q4_0_4_4/4_8/8_8, and these would have their weights interleaved in memory in order to improve performance on ARM and AVX machines by loading up more data in one pass.
Now, however, there is something called "online repacking" for weights. details in [this PR](https://github.com/ggerganov/llama.cpp/pull/9921). If you use Q4_0 and your hardware would benefit from repacking weights, it will do it automatically on the fly.
As of llama.cpp build [b4282](https://github.com/ggerganov/llama.cpp/releases/tag/b4282) you will not be able to run the Q4_0_X_X files and will instead need to use Q4_0.
Additionally, if you want to get slightly better quality for , you can use IQ4_NL thanks to [this PR](https://github.com/ggerganov/llama.cpp/pull/10541) which will also repack the weights for ARM, though only the 4_4 for now. The loading time may be slower but it will result in an overall speed incrase.
<details>
<summary>Click to view Q4_0_X_X information (deprecated</summary>
I'm keeping this section to show the potential theoretical uplift in performance from using the Q4_0 with online repacking.
<details>
<summary>Click to view benchmarks on an AVX2 system (EPYC7702)</summary>
| model | size | params | backend | threads | test | t/s | % (vs Q4_0) |
| ------------------------------ | ---------: | ---------: | ---------- | ------: | ------------: | -------------------: |-------------: |
| qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | pp512 | 204.03 ± 1.03 | 100% |
| qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | pp1024 | 282.92 ± 0.19 | 100% |
| qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | pp2048 | 259.49 ± 0.44 | 100% |
| qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | tg128 | 39.12 ± 0.27 | 100% |
| qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | tg256 | 39.31 ± 0.69 | 100% |
| qwen2 3B Q4_0 | 1.70 GiB | 3.09 B | CPU | 64 | tg512 | 40.52 ± 0.03 | 100% |
| qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | pp512 | 301.02 ± 1.74 | 147% |
| qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | pp1024 | 287.23 ± 0.20 | 101% |
| qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | pp2048 | 262.77 ± 1.81 | 101% |
| qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | tg128 | 18.80 ± 0.99 | 48% |
| qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | tg256 | 24.46 ± 3.04 | 83% |
| qwen2 3B Q4_K_M | 1.79 GiB | 3.09 B | CPU | 64 | tg512 | 36.32 ± 3.59 | 90% |
| qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | pp512 | 271.71 ± 3.53 | 133% |
| qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | pp1024 | 279.86 ± 45.63 | 100% |
| qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | pp2048 | 320.77 ± 5.00 | 124% |
| qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | tg128 | 43.51 ± 0.05 | 111% |
| qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | tg256 | 43.35 ± 0.09 | 110% |
| qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | tg512 | 42.60 ± 0.31 | 105% |
Q4_0_8_8 offers a nice bump to prompt processing and a small bump to text generation
</details>
</details>
## Which file should I choose?
<details>
<summary>Click here for details</summary>
A great write up with charts showing various performances is provided by Artefact2 [here](https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9)
The first thing to figure out is how big a model you can run. To do this, you'll need to figure out how much RAM and/or VRAM you have.
If you want your model running as FAST as possible, you'll want to fit the whole thing on your GPU's VRAM. Aim for a quant with a file size 1-2GB smaller than your GPU's total VRAM.
If you want the absolute maximum quality, add both your system RAM and your GPU's VRAM together, then similarly grab a quant with a file size 1-2GB Smaller than that total.
Next, you'll need to decide if you want to use an 'I-quant' or a 'K-quant'.
If you don't want to think too much, grab one of the K-quants. These are in format 'QX_K_X', like Q5_K_M.
If you want to get more into the weeds, you can check out this extremely useful feature chart:
[llama.cpp feature matrix](https://github.com/ggerganov/llama.cpp/wiki/Feature-matrix)
But basically, if you're aiming for below Q4, and you're running cuBLAS (Nvidia) or rocBLAS (AMD), you should look towards the I-quants. These are in format IQX_X, like IQ3_M. These are newer and offer better performance for their size.
These I-quants can also be used on CPU and Apple Metal, but will be slower than their K-quant equivalent, so speed vs performance is a tradeoff you'll have to decide.
The I-quants are *not* compatible with Vulcan, which is also AMD, so if you have an AMD card double check if you're using the rocBLAS build or the Vulcan build. At the time of writing this, LM Studio has a preview with ROCm support, and other inference engines have specific builds for ROCm.
</details>
## Credits
Thank you kalomaze and Dampf for assistance in creating the imatrix calibration dataset.
Thank you ZeroWw for the inspiration to experiment with embed/output.
Thank you to LM Studio for sponsoring my work.
Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski

3
ibm-granite_granite-3.2-2b-instruct.imatrix Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:381ecc0cd03fbbb51284952e0c0654cf5d3ac65d98f6081f26cdacffc55c1db4
size 3286098