Initial GGML model commit

README.md

@@ -28,27 +28,60 @@ GGML files are for CPU + GPU inference using [llama.cpp](https://github.com/gger
 * [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)
 * [ctransformers](https://github.com/marella/ctransformers)
 
-## Other repositories available
+## Repositories available
 
-* [4bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/Wizard-Vicuna-30B-Uncensored-GPTQ).
-* [4bit and 5bit GGML models for CPU inference](https://huggingface.co/TheBloke/Wizard-Vicuna-30B-Uncensored-GGML).
-* [float16 HF format model for GPU inference and further conversions](https://huggingface.co/TheBloke/Wizard-Vicuna-30B-Uncensored-fp16).
-*
-## THE FILES IN MAIN BRANCH REQUIRES LATEST LLAMA.CPP (May 19th 2023 - commit 2d5db48)!
+* [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/Wizard-Vicuna-30B-Uncensored-GGML)
+* [2, 3, 4, 5, 6 and 8-bit GGML models for CPU+GPU inference](https://huggingface.co/TheBloke/Wizard-Vicuna-30B-Uncensored-GGML)
+* [Unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/ehartford/Wizard-Vicuna-30B-Uncensored)
 
-llama.cpp recently made another breaking change to its quantisation methods - https://github.com/ggerganov/llama.cpp/pull/1508
+<!-- compatibility_ggml start -->
+## Compatibility
 
-I have quantised the GGML files in this repo with the latest version. Therefore you will require llama.cpp compiled on May 19th or later (commit `2d5db48` or later) to use them.
+### Original llama.cpp quant methods: `q4_0, q4_1, q5_0, q5_1, q8_0`
+
+I have quantized these 'original' quantisation methods using an older version of llama.cpp so that they remain compatible with llama.cpp as of May 19th, commit `2d5db48`.
+
+They should be compatible with all current UIs and libraries that use llama.cpp, such as those listed at the top of this README.
+
+### New k-quant methods: `q2_K, q3_K_S, q3_K_M, q3_K_L, q4_K_S, q4_K_M, q5_K_S, q6_K`
+
+These new quantisation methods are only compatible with llama.cpp as of June 6th, commit `2d43387`.
+
+They will NOT be compatible with koboldcpp, text-generation-ui, and other UIs and libraries yet. Support is expected to come over the next few days.
+
+## Explanation of the new k-quant methods
+
+The new methods available are:
+* GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
+* GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
+* GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
+* GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
+* GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
+* GGML_TYPE_Q8_K - "type-0" 8-bit quantization. Only used for quantizing intermediate results. The difference to the existing Q8_0 is that the block size is 256. All 2-6 bit dot products are implemented for this quantization type.
+
+Refer to the Provided Files table below to see what files use which methods, and how.
+<!-- compatibility_ggml end -->
 
 ## Provided files
-| Name | Quant method | Bits | Size | RAM required | Use case |
+| Name | Quant method | Bits | Size | Max RAM required | Use case |
 | ---- | ---- | ---- | ---- | ---- | ----- |
-| Wizard-Vicuna-30B-Uncensored.ggmlv3.q4_0.bin | q4_0 | 4 | 18.30 GB | 20.80 GB | 4-bit. |
-| Wizard-Vicuna-30B-Uncensored.ggmlv3.q4_1.bin | q4_1 | 4 | 20.33 GB | 22.83 GB | 4-bit. Higher accuracy than q4_0 but not as high as q5_0. However has quicker inference than q5 models. |
-| Wizard-Vicuna-30B-Uncensored.ggmlv3.q5_0.bin | q5_0 | 5 | 22.37 GB | 24.87 GB | 5-bit. Higher accuracy, higher resource usage and slower inference. |
-| Wizard-Vicuna-30B-Uncensored.ggmlv3.q5_1.bin | q5_1 | 5 | 24.40 GB | 26.90 GB | 5-bit. Even higher accuracy, resource usage and slower inference. |
-| Wizard-Vicuna-30B-Uncensored.ggmlv3.q8_0.bin | q8_0 | 8 | 34.56 GB | 37.06 GB | 8-bit. Almost indistinguishable from float16. Huge resource use and slow. Not recommended for normal use. |
-
+| Wizard-Vicuna-30B-Uncensored.ggmlv3.q2_K.bin | q2_K | 2 | 13.60 GB | 16.10 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.vw and feed_forward.w2 tensors, GGML_TYPE_Q2_K for the other tensors. |
+| Wizard-Vicuna-30B-Uncensored.ggmlv3.q3_K_L.bin | q3_K_L | 3 | 17.20 GB | 19.70 GB | New k-quant method. Uses GGML_TYPE_Q5_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
+| Wizard-Vicuna-30B-Uncensored.ggmlv3.q3_K_M.bin | q3_K_M | 3 | 15.64 GB | 18.14 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
+| Wizard-Vicuna-30B-Uncensored.ggmlv3.q3_K_S.bin | q3_K_S | 3 | 13.98 GB | 16.48 GB | New k-quant method. Uses GGML_TYPE_Q3_K for all tensors |
+| Wizard-Vicuna-30B-Uncensored.ggmlv3.q4_0.bin | q4_0 | 4 | 18.30 GB | 20.80 GB | Original llama.cpp quant method, 4-bit. |
+| Wizard-Vicuna-30B-Uncensored.ggmlv3.q4_1.bin | q4_1 | 4 | 20.33 GB | 22.83 GB | Original llama.cpp quant method, 4-bit. Higher accuracy than q4_0 but not as high as q5_0. However has quicker inference than q5 models. |
+| Wizard-Vicuna-30B-Uncensored.ggmlv3.q4_K_M.bin | q4_K_M | 4 | 19.57 GB | 22.07 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q4_K |
+| Wizard-Vicuna-30B-Uncensored.ggmlv3.q4_K_S.bin | q4_K_S | 4 | 18.30 GB | 20.80 GB | New k-quant method. Uses GGML_TYPE_Q4_K for all tensors |
+| Wizard-Vicuna-30B-Uncensored.ggmlv3.q5_0.bin | q5_0 | 5 | 22.37 GB | 24.87 GB | Original llama.cpp quant method, 5-bit. Higher accuracy, higher resource usage and slower inference. |
+| Wizard-Vicuna-30B-Uncensored.ggmlv3.q5_1.bin | q5_1 | 5 | 24.40 GB | 26.90 GB | Original llama.cpp quant method, 5-bit. Even higher accuracy, resource usage and slower inference. |
+| Wizard-Vicuna-30B-Uncensored.ggmlv3.q5_K_M.bin | q5_K_M | 5 | 23.02 GB | 25.52 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q5_K |
+| Wizard-Vicuna-30B-Uncensored.ggmlv3.q5_K_S.bin | q5_K_S | 5 | 22.37 GB | 24.87 GB | New k-quant method. Uses GGML_TYPE_Q5_K for all tensors |
+| Wizard-Vicuna-30B-Uncensored.ggmlv3.q6_K.bin | q6_K | 6 | 26.69 GB | 29.19 GB | New k-quant method. Uses GGML_TYPE_Q8_K - 6-bit quantization - for all tensors |
+| Wizard-Vicuna-30B-Uncensored.ggmlv3.q8_0.bin | q8_0 | 8 | 34.56 GB | 37.06 GB | Original llama.cpp quant method, 8-bit. Almost indistinguishable from float16. High resource use and slow. Not recommended for most users. |
+
+
+**Note**: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.
 
 ## How to run in `llama.cpp`
 
@@ -87,20 +120,24 @@ Donaters will get priority support on any and all AI/LLM/model questions and req
 * Patreon: https://patreon.com/TheBlokeAI
 * Ko-Fi: https://ko-fi.com/TheBlokeAI
 
-**Patreon special mentions**: Aemon Algiz, Dmitriy Samsonov, Nathan LeClaire, Trenton Dambrowitz, Mano Prime, David Flickinger, vamX, Nikolai Manek, senxiiz, Khalefa Al-Ahmad, Illia Dulskyi, Jonathan Leane, Talal Aujan, V. Lukas, Joseph William Delisle, Pyrater, Oscar Rangel, Lone Striker, Luke Pendergrass, Eugene Pentland, Sebastain Graf, Johann-Peter Hartman.
+**Special thanks to**: Luke from CarbonQuill, Aemon Algiz, Dmitriy Samsonov.
+
+**Patreon special mentions**: Ajan Kanaga, Kalila, Derek Yates, Sean Connelly, Luke, Nathan LeClaire, Trenton Dambrowitz, Mano Prime, David Flickinger, vamX, Nikolai Manek, senxiiz, Khalefa Al-Ahmad, Illia Dulskyi, trip7s trip, Jonathan Leane, Talal Aujan, Artur Olbinski, Cory Kujawski, Joseph William Delisle, Pyrater, Oscar Rangel, Lone Striker, Luke Pendergrass, Eugene Pentland, Johann-Peter Hartmann.
 
 Thank you to all my generous patrons and donaters!
+
 <!-- footer end -->
 
 # Original model card: Eric Hartford's Wizard Vicuna 30B Uncensored
 
+
 This is [wizard-vicuna-13b](https://huggingface.co/junelee/wizard-vicuna-13b) trained with a subset of the dataset - responses that contained alignment / moralizing were removed. The intent is to train a WizardLM that doesn't have alignment built-in, so that alignment (of any sort) can be added separately with for example with a RLHF LoRA.
 
 Shout out to the open source AI/ML community, and everyone who helped me out.
 
-Note:
+Note:  
 
-An uncensored model has no guardrails.
+An uncensored model has no guardrails.  
 
 You are responsible for anything you do with the model, just as you are responsible for anything you do with any dangerous object such as a knife, gun, lighter, or car.