--- base_model: WizardLM/WizardLM-30B-V1.0 language: - en library_name: transformers no_imatrix: 'GGML_ASSERT: llama.cpp/ggml-quants.c:12166: besti1 >= 0 && besti2 >= 0 && best_k >= 0' quantized_by: mradermacher --- ## About weighted/imatrix quants of https://huggingface.co/WizardLM/WizardLM-30B-V1.0 **No IQ1\* quants as llama.cpp is crashing when trying to generate it** static quants are available at https://huggingface.co/mradermacher/WizardLM-30B-V1.0-GGUF ## Usage If you are unsure how to use GGUF files, refer to one of [TheBloke's READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for more details, including on how to concatenate multi-part files. ## Provided Quants (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants) | Link | Type | Size/GB | Notes | |:-----|:-----|--------:|:------| | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 8.8 | | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-IQ2_XS.gguf) | i1-IQ2_XS | 9.7 | | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-IQ2_S.gguf) | i1-IQ2_S | 10.5 | | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-IQ2_M.gguf) | i1-IQ2_M | 11.3 | | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-Q2_K.gguf) | i1-Q2_K | 12.1 | IQ3_XXS probably better | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 12.4 | lower quality | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-IQ3_XS.gguf) | i1-IQ3_XS | 13.4 | | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-IQ3_S.gguf) | i1-IQ3_S | 14.2 | beats Q3_K* | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-Q3_K_S.gguf) | i1-Q3_K_S | 14.2 | IQ3_XS probably better | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-IQ3_M.gguf) | i1-IQ3_M | 15.0 | | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-Q3_K_M.gguf) | i1-Q3_K_M | 15.9 | IQ3_S probably better | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-Q3_K_L.gguf) | i1-Q3_K_L | 17.4 | IQ3_M probably better | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-IQ4_XS.gguf) | i1-IQ4_XS | 17.4 | | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-Q4_0.gguf) | i1-Q4_0 | 18.5 | fast, low quality | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-Q4_K_S.gguf) | i1-Q4_K_S | 18.6 | optimal size/speed/quality | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-Q4_K_M.gguf) | i1-Q4_K_M | 19.7 | fast, recommended | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-Q5_K_S.gguf) | i1-Q5_K_S | 22.5 | | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-Q5_K_M.gguf) | i1-Q5_K_M | 23.1 | | | [GGUF](https://huggingface.co/mradermacher/WizardLM-30B-V1.0-i1-GGUF/resolve/main/WizardLM-30B-V1.0.i1-Q6_K.gguf) | i1-Q6_K | 26.8 | practically like static Q6_K | Here is a handy graph by ikawrakow comparing some lower-quality quant types (lower is better): ![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png) And here are Artefact2's thoughts on the matter: https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9 ## FAQ / Model Request See https://huggingface.co/mradermacher/model_requests for some answers to questions you might have and/or if you want some other model quantized. ## Thanks I thank my company, [nethype GmbH](https://www.nethype.de/), for letting me use its servers and providing upgrades to my workstation to enable this work in my free time.