Model Overview
Model Features
Model Capabilities
Use Cases
🚀 Llamacpp imatrix Quantizations of c4ai-command-r7b-12-2024-abliterated
This project provides Llama.cpp imatrix quantizations of the c4ai-command-r7b-12-2024-abliterated model. It offers various quantization types to balance between model quality and resource usage, enabling users to run the model efficiently on different hardware.
Metadata
Property | Details |
---|---|
Quantized By | bartowski |
Pipeline Tag | text-generation |
Extra Gated Fields | Name: text, Affiliation: text, Country: country, I agree to use this model for non-commercial use ONLY: checkbox |
Base Model | huihui-ai/c4ai-command-r7b-12-2024-abliterated |
License | cc-by-nc-4.0 |
Tags | abliterated, uncensored |
Languages | en, fr, de, es, it, pt, ja, ko, zh, ar, el, fa, pl, id, cs, he, hi, nl, ro, ru, tr, uk, vi |
Inference | false |
Extra Gated Prompt | By submitting this form, you agree to the License Agreement and acknowledge that the information you provide will be collected, used, and shared in accordance with Cohere’s Privacy Policy. You’ll receive email updates about C4AI and Cohere research, events, products and services. You can unsubscribe at any time. |
🚀 Quick Start
Quantization
Using llama.cpp release b4415 for quantization. All quants are made using the imatrix option with the dataset from here.
Running the Model
You can run the quantized models in LM Studio.
✨ Features
Prompt Format
<|START_OF_TURN_TOKEN|><|SYSTEM_TOKEN|>{system_prompt}<|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|USER_TOKEN|>{prompt}<|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|><|START_RESPONSE|><|END_RESPONSE|><|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|><|START_RESPONSE|>
Download Options
You can download a specific file from the table below:
Filename | Quant type | File Size | Split | Description |
---|---|---|---|---|
c4ai-command-r7b-12-2024-abliterated-bf16.gguf | bf16 | 16.07GB | false | Full BF16 weights. |
c4ai-command-r7b-12-2024-abliterated-Q8_0.gguf | Q8_0 | 8.54GB | false | Extremely high quality, generally unneeded but max available quant. |
c4ai-command-r7b-12-2024-abliterated-Q6_K_L.gguf | Q6_K_L | 6.85GB | false | Uses Q8_0 for embed and output weights. Very high quality, near perfect, recommended. |
c4ai-command-r7b-12-2024-abliterated-Q6_K.gguf | Q6_K | 6.60GB | false | Very high quality, near perfect, recommended. |
c4ai-command-r7b-12-2024-abliterated-Q5_K_L.gguf | Q5_K_L | 6.06GB | false | Uses Q8_0 for embed and output weights. High quality, recommended. |
c4ai-command-r7b-12-2024-abliterated-Q5_K_M.gguf | Q5_K_M | 5.80GB | false | High quality, recommended. |
c4ai-command-r7b-12-2024-abliterated-Q5_K_S.gguf | Q5_K_S | 5.67GB | false | High quality, recommended. |
c4ai-command-r7b-12-2024-abliterated-Q4_K_L.gguf | Q4_K_L | 5.31GB | false | Uses Q8_0 for embed and output weights. Good quality, recommended. |
c4ai-command-r7b-12-2024-abliterated-Q4_1.gguf | Q4_1 | 5.23GB | false | Legacy format, similar performance to Q4_K_S but with improved tokens/watt on Apple silicon. |
c4ai-command-r7b-12-2024-abliterated-Q4_K_M.gguf | Q4_K_M | 5.06GB | false | Good quality, default size for most use cases, recommended. |
c4ai-command-r7b-12-2024-abliterated-Q4_K_S.gguf | Q4_K_S | 4.83GB | false | Slightly lower quality with more space savings, recommended. |
c4ai-command-r7b-12-2024-abliterated-Q4_0.gguf | Q4_0 | 4.81GB | false | Legacy format, offers online repacking for ARM and AVX CPU inference. |
c4ai-command-r7b-12-2024-abliterated-IQ4_NL.gguf | IQ4_NL | 4.81GB | false | Similar to IQ4_XS, but slightly larger. Offers online repacking for ARM CPU inference. |
c4ai-command-r7b-12-2024-abliterated-Q3_K_XL.gguf | Q3_K_XL | 4.78GB | false | Uses Q8_0 for embed and output weights. Lower quality but usable, good for low RAM availability. |
c4ai-command-r7b-12-2024-abliterated-IQ4_XS.gguf | IQ4_XS | 4.60GB | false | Decent quality, smaller than Q4_K_S with similar performance, recommended. |
c4ai-command-r7b-12-2024-abliterated-Q3_K_L.gguf | Q3_K_L | 4.53GB | false | Lower quality but usable, good for low RAM availability. |
c4ai-command-r7b-12-2024-abliterated-Q3_K_M.gguf | Q3_K_M | 4.22GB | false | Low quality. |
c4ai-command-r7b-12-2024-abliterated-IQ3_M.gguf | IQ3_M | 3.99GB | false | Medium-low quality, new method with decent performance comparable to Q3_K_M. |
c4ai-command-r7b-12-2024-abliterated-Q3_K_S.gguf | Q3_K_S | 3.87GB | false | Low quality, not recommended. |
c4ai-command-r7b-12-2024-abliterated-IQ3_XS.gguf | IQ3_XS | 3.72GB | false | Lower quality, new method with decent performance, slightly better than Q3_K_S. |
c4ai-command-r7b-12-2024-abliterated-Q2_K_L.gguf | Q2_K_L | 3.69GB | false | Uses Q8_0 for embed and output weights. Very low quality but surprisingly usable. |
c4ai-command-r7b-12-2024-abliterated-Q2_K.gguf | Q2_K | 3.44GB | false | Very low quality but surprisingly usable. |
c4ai-command-r7b-12-2024-abliterated-IQ2_M.gguf | IQ2_M | 3.08GB | false | Relatively low quality, uses SOTA techniques to be surprisingly usable. |
📦 Installation
Downloading using huggingface-cli
Click to view download instructions
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/c4ai-command-r7b-12-2024-abliterated-GGUF --include "c4ai-command-r7b-12-2024-abliterated-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/c4ai-command-r7b-12-2024-abliterated-GGUF --include "c4ai-command-r7b-12-2024-abliterated-Q8_0/*" --local-dir ./
You can either specify a new local-dir (c4ai-command-r7b-12-2024-abliterated-Q8_0) or download them all in place (./)
🔧 Technical Details
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.
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. 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 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, you can use IQ4_NL thanks to this PR 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 increase.
Click to view Q4_0_X_X information (deprecated)
I'm keeping this section to show the potential theoretical uplift in performance from using the Q4_0 with online repacking.
Click to view benchmarks on an AVX2 system (EPYC7702)
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 | 39.12 ± 0.27 | 100% |
qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | tg256 | 39.31 ± 0.69 | 100% |
qwen2 3B Q4_0_8_8 | 1.69 GiB | 3.09 B | CPU | 64 | tg512 | 40.52 ± 0.03 | 100% |
📄 License
This project is licensed under the cc-by-nc-4.0 license.

