bartowski/OpenCrystal-12B-L3-GGUF
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: bartowski/OpenCrystal-12B-L3-GGUF
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-08-21 19:55:13 +08:00
commit 5e4e4a01d9
29 changed files with 362 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

61
.gitattributes vendored Normal file
View File

@@ -0,0 +1,61 @@
*.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
OpenCrystal-12B-L3-Q8_0.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q6_K_L.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q6_K.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q5_K_L.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q5_K_M.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q5_K_S.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q4_K_L.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q4_K_M.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q4_K_S.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q4_0_8_8.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q4_0_4_8.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q4_0_4_4.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q4_0.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-IQ4_XS.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q3_K_XL.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q3_K_L.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q3_K_M.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-IQ3_M.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q3_K_S.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-IQ3_XS.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q2_K_L.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-Q2_K.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-IQ2_M.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-IQ2_S.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3-f16.gguf filter=lfs diff=lfs merge=lfs -text
OpenCrystal-12B-L3.imatrix filter=lfs diff=lfs merge=lfs -text

3
OpenCrystal-12B-L3-IQ2_M.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:6cf661cb167eabbdec2a534613b1e8b1fbb8f01f6b483eebd227baf4b9b4d760
size 4125054240

3
OpenCrystal-12B-L3-IQ2_S.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:0b5643130ce732229b4fb62a8f3f3311ca6bcbe25a062731ebb667b0ca22b5fe
size 3840365856

3
OpenCrystal-12B-L3-IQ3_M.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:84bae195b3d67056ea3ad13913ca1fd26fefb659ed5d19e887a6f4d923febe75
size 5344982304

3
OpenCrystal-12B-L3-IQ3_XS.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:865f2602f9c86fb8cdcbb95b37277c9cd5b004be50f90c46fd6696881e68f169
size 4945868064

3
OpenCrystal-12B-L3-IQ4_XS.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:f44372e73b5623d040647cffcd201e1ba7fe031f66804c4e78d34f1f0cb64f90
size 6312564000

3
OpenCrystal-12B-L3-Q2_K.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:f86841d0cf7880bfb82f04c1178c3e4b40a1e78fbe42886c259fd7f680abc250
size 4463121696

3
OpenCrystal-12B-L3-Q2_K_L.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:105b0352594d0371f3c9c3fa72507821d12a2a8330bf3684c47c4ad6a99f1273
size 4976145696

3
OpenCrystal-12B-L3-Q3_K_L.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:4d312160927f77cd39465906e0bf530508eedf17490fb0c01f86424d2102ce95
size 6150681888

3
OpenCrystal-12B-L3-Q3_K_M.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:1aa552fa3c87c1311fcf401906e8730d81f56ff64b7f033e6bfeafbd7e58c15b
size 5695599904

3
OpenCrystal-12B-L3-Q3_K_S.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:9082f978eeeca1dcc915db0ef05fe2d40f4de758eab4f4c4d7722384adb10dd8
size 5164496160

3
OpenCrystal-12B-L3-Q3_K_XL.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:4d7a02a93fa1df522135d926e48a964fa09b21852ccd9cefbd7c3f0b2c155f7d
size 6610351392

3
OpenCrystal-12B-L3-Q4_0.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:9e2906a6215f56fd766da88648f3724e76da6ef645b52bc331b0565bbf8ab16b
size 6646699296

3
OpenCrystal-12B-L3-Q4_0_4_4.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:76f26147fcbfcbc3fd4f45c1f0fc312696ec955ee189d22ef3ed3ec2fdbf8d2a
size 6624679200

3
OpenCrystal-12B-L3-Q4_0_4_8.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:8bfd96a57bca1e566df823ca963bc44aa1485272625e56c4ef39548b13210076
size 6624679200

3
OpenCrystal-12B-L3-Q4_0_8_8.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:426b47d871e5a843e45a14d25f3d67607ba3ff2ee6945bac8e8b383c304b6009
size 6624679200

3
OpenCrystal-12B-L3-Q4_K_L.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:4177a4ad92a5f09026aaa42e103e0a226aa671bacbacfe89e30efb550d827fed
size 7403861280

3
OpenCrystal-12B-L3-Q4_K_M.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:a89f064beacee71d297eda65cb6770a3873cdcec50df6fdc03fd6d90c424c180
size 7013963040

3
OpenCrystal-12B-L3-Q4_K_S.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:5a7d9f0f880fb3b7ee95c2018d439705212f6cdcff768cd11f5981d84d68ed67
size 6670816544

3
OpenCrystal-12B-L3-Q5_K_L.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:9842c8765d46db9683d16dcacfef3e2bda7c61de52168c859b13f8c2fc0d60fe
size 8523740448

3
OpenCrystal-12B-L3-Q5_K_M.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:e2f06a4754279a1eaa44211c839a5a0282c0510ad4bd51430ac24ad1d6d59fd0
size 8199509280

3
OpenCrystal-12B-L3-Q5_K_S.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:a5ff236151bbf24b51fe15aa8c5150eee252b8ea2c31273015795b00d7c2ff44
size 7998969120

3
OpenCrystal-12B-L3-Q6_K.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:699967ff5e896cf5d2204d908d7cde2967ceb5659598af825bf574f62ff14a5a
size 9459152160

3
OpenCrystal-12B-L3-Q6_K_L.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:b56d460705f3563384dfb1a1b8e881056bf7ae732742a04561931cc5980a312d
size 9713612064

3
OpenCrystal-12B-L3-Q8_0.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:a14d38b1c072a441ad462f3b3b75edef8b9657bf7396e8b9931d5c0adb8ac207
size 12249068832

3
OpenCrystal-12B-L3-f16.gguf Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:0ddf7e4bde084ebf10479e5fadb207805bbb7df10ab45d7347c8670a9717ac47
size 23048745984

3
OpenCrystal-12B-L3.imatrix Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:f87ce3fda85362dc501a7b726e33fd804cbe816b6ee856435dbe3d55bb764431
size 7482266

222
README.md Normal file
View File

@@ -0,0 +1,222 @@
---
base_model: Darkknight535/OpenCrystal-12B-L3
pipeline_tag: text-generation
tags:
- not-for-all-audiences
quantized_by: bartowski
model-index:
- name: OpenCrystal-12B-L3
results:
- task:
type: text-generation
name: Text Generation
dataset:
name: IFEval (0-Shot)
type: HuggingFaceH4/ifeval
args:
num_few_shot: 0
metrics:
- type: inst_level_strict_acc and prompt_level_strict_acc
value: 40.71
name: strict accuracy
source:
url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Darkknight535/OpenCrystal-12B-L3
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: BBH (3-Shot)
type: BBH
args:
num_few_shot: 3
metrics:
- type: acc_norm
value: 31.84
name: normalized accuracy
source:
url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Darkknight535/OpenCrystal-12B-L3
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: MATH Lvl 5 (4-Shot)
type: hendrycks/competition_math
args:
num_few_shot: 4
metrics:
- type: exact_match
value: 7.93
name: exact match
source:
url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Darkknight535/OpenCrystal-12B-L3
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: GPQA (0-shot)
type: Idavidrein/gpqa
args:
num_few_shot: 0
metrics:
- type: acc_norm
value: 7.49
name: acc_norm
source:
url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Darkknight535/OpenCrystal-12B-L3
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: MuSR (0-shot)
type: TAUR-Lab/MuSR
args:
num_few_shot: 0
metrics:
- type: acc_norm
value: 5.74
name: acc_norm
source:
url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Darkknight535/OpenCrystal-12B-L3
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: MMLU-PRO (5-shot)
type: TIGER-Lab/MMLU-Pro
config: main
split: test
args:
num_few_shot: 5
metrics:
- type: acc
value: 29.34
name: accuracy
source:
url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Darkknight535/OpenCrystal-12B-L3
name: Open LLM Leaderboard
---
## Llamacpp imatrix Quantizations of OpenCrystal-12B-L3
Using <a href="https://github.com/ggerganov/llama.cpp/">llama.cpp</a> release <a href="https://github.com/ggerganov/llama.cpp/releases/tag/b3878">b3878</a> for quantization.
Original model: https://huggingface.co/Darkknight535/OpenCrystal-12B-L3
All quants made using imatrix option with dataset from [here](https://gist.github.com/bartowski1182/eb213dccb3571f863da82e99418f81e8)
Run them in [LM Studio](https://lmstudio.ai/)
## Prompt format
```
<|begin_of_text|><|start_header_id|>system<|end_header_id|>
{system_prompt}<|eot_id|><|start_header_id|>user<|end_header_id|>
{prompt}<|eot_id|><|start_header_id|>assistant<|end_header_id|>
```
## Download a file (not the whole branch) from below:
| Filename | Quant type | File Size | Split | Description |
| -------- | ---------- | --------- | ----- | ----------- |
| [OpenCrystal-12B-L3-f16.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-f16.gguf) | f16 | 23.05GB | false | Full F16 weights. |
| [OpenCrystal-12B-L3-Q8_0.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q8_0.gguf) | Q8_0 | 12.25GB | false | Extremely high quality, generally unneeded but max available quant. |
| [OpenCrystal-12B-L3-Q6_K_L.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q6_K_L.gguf) | Q6_K_L | 9.71GB | false | Uses Q8_0 for embed and output weights. Very high quality, near perfect, *recommended*. |
| [OpenCrystal-12B-L3-Q6_K.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q6_K.gguf) | Q6_K | 9.46GB | false | Very high quality, near perfect, *recommended*. |
| [OpenCrystal-12B-L3-Q5_K_L.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q5_K_L.gguf) | Q5_K_L | 8.52GB | false | Uses Q8_0 for embed and output weights. High quality, *recommended*. |
| [OpenCrystal-12B-L3-Q5_K_M.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q5_K_M.gguf) | Q5_K_M | 8.20GB | false | High quality, *recommended*. |
| [OpenCrystal-12B-L3-Q5_K_S.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q5_K_S.gguf) | Q5_K_S | 8.00GB | false | High quality, *recommended*. |
| [OpenCrystal-12B-L3-Q4_K_L.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q4_K_L.gguf) | Q4_K_L | 7.40GB | false | Uses Q8_0 for embed and output weights. Good quality, *recommended*. |
| [OpenCrystal-12B-L3-Q4_K_M.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q4_K_M.gguf) | Q4_K_M | 7.01GB | false | Good quality, default size for must use cases, *recommended*. |
| [OpenCrystal-12B-L3-Q4_K_S.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q4_K_S.gguf) | Q4_K_S | 6.67GB | false | Slightly lower quality with more space savings, *recommended*. |
| [OpenCrystal-12B-L3-Q4_0.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q4_0.gguf) | Q4_0 | 6.65GB | false | Legacy format, generally not worth using over similarly sized formats |
| [OpenCrystal-12B-L3-Q4_0_8_8.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q4_0_8_8.gguf) | Q4_0_8_8 | 6.62GB | false | Optimized for ARM inference. Requires 'sve' support (see link below). *Don't use on Mac or Windows*. |
| [OpenCrystal-12B-L3-Q4_0_4_8.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q4_0_4_8.gguf) | Q4_0_4_8 | 6.62GB | false | Optimized for ARM inference. Requires 'i8mm' support (see link below). *Don't use on Mac or Windows*. |
| [OpenCrystal-12B-L3-Q4_0_4_4.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q4_0_4_4.gguf) | Q4_0_4_4 | 6.62GB | false | Optimized for ARM inference. Should work well on all ARM chips, pick this if you're unsure. *Don't use on Mac or Windows*. |
| [OpenCrystal-12B-L3-Q3_K_XL.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q3_K_XL.gguf) | Q3_K_XL | 6.61GB | false | Uses Q8_0 for embed and output weights. Lower quality but usable, good for low RAM availability. |
| [OpenCrystal-12B-L3-IQ4_XS.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-IQ4_XS.gguf) | IQ4_XS | 6.31GB | false | Decent quality, smaller than Q4_K_S with similar performance, *recommended*. |
| [OpenCrystal-12B-L3-Q3_K_L.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q3_K_L.gguf) | Q3_K_L | 6.15GB | false | Lower quality but usable, good for low RAM availability. |
| [OpenCrystal-12B-L3-Q3_K_M.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q3_K_M.gguf) | Q3_K_M | 5.70GB | false | Low quality. |
| [OpenCrystal-12B-L3-IQ3_M.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-IQ3_M.gguf) | IQ3_M | 5.34GB | false | Medium-low quality, new method with decent performance comparable to Q3_K_M. |
| [OpenCrystal-12B-L3-Q3_K_S.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q3_K_S.gguf) | Q3_K_S | 5.16GB | false | Low quality, not recommended. |
| [OpenCrystal-12B-L3-Q2_K_L.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q2_K_L.gguf) | Q2_K_L | 4.98GB | false | Uses Q8_0 for embed and output weights. Very low quality but surprisingly usable. |
| [OpenCrystal-12B-L3-IQ3_XS.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-IQ3_XS.gguf) | IQ3_XS | 4.95GB | false | Lower quality, new method with decent performance, slightly better than Q3_K_S. |
| [OpenCrystal-12B-L3-Q2_K.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-Q2_K.gguf) | Q2_K | 4.46GB | false | Very low quality but surprisingly usable. |
| [OpenCrystal-12B-L3-IQ2_M.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-IQ2_M.gguf) | IQ2_M | 4.13GB | false | Relatively low quality, uses SOTA techniques to be surprisingly usable. |
| [OpenCrystal-12B-L3-IQ2_S.gguf](https://huggingface.co/bartowski/OpenCrystal-12B-L3-GGUF/blob/main/OpenCrystal-12B-L3-IQ2_S.gguf) | IQ2_S | 3.84GB | false | Low quality, uses SOTA techniques to be 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.
Some say that this improves the quality, others don't notice any difference. If you use these models PLEASE COMMENT with your findings. I would like feedback that these are actually used and useful so I don't keep uploading quants no one is using.
Thanks!
## Downloading using huggingface-cli
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/OpenCrystal-12B-L3-GGUF --include "OpenCrystal-12B-L3-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/OpenCrystal-12B-L3-GGUF --include "OpenCrystal-12B-L3-Q8_0/*" --local-dir ./
```
You can either specify a new local-dir (OpenCrystal-12B-L3-Q8_0) or download them all in place (./)
## Q4_0_X_X
These are *NOT* for Metal (Apple) offloading, only ARM chips.
If you're using an ARM chip, the Q4_0_X_X quants will have a substantial speedup. Check out Q4_0_4_4 speed comparisons [on the original pull request](https://github.com/ggerganov/llama.cpp/pull/5780#pullrequestreview-21657544660)
To check which one would work best for your ARM chip, you can check [AArch64 SoC features](https://gpages.juszkiewicz.com.pl/arm-socs-table/arm-socs.html) (thanks EloyOn!).
## Which file should I choose?
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.
## 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
Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski

1
configuration.json Normal file
View File

@@ -0,0 +1 @@
{"framework": "pytorch", "task": "text-generation", "allow_remote": true}