arcee-ai
/

SuperNova-Medius-GGUF

@@ -1,129 +1,68 @@
 ---
-base_model: arcee-train/Qwen2.5-14B-LlamaDistill-Merged-v9
-pipeline_tag: text-generation
 tags:
 - mergekit
 - merge
-quantized_by: bartowski
 ---
-## Llamacpp imatrix Quantizations of SuperNova-14B
-Using <a href="https://github.com/ggerganov/llama.cpp/">llama.cpp</a> release <a href="https://github.com/ggerganov/llama.cpp/releases/tag/b3901">b3901</a> for quantization.
-Original model: https://huggingface.co/arcee-train/Qwen2.5-14B-LlamaDistill-Merged-v9
-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
-```
-<|im_start|>system
-{system_prompt}<|im_end|>
-<|im_start|>user
-{prompt}<|im_end|>
-<|im_start|>assistant
-```
-## Download a file (not the whole branch) from below:
-| Filename | Quant type | File Size | Split | Description |
-| -------- | ---------- | --------- | ----- | ----------- |
-| [SuperNova-14B-f16.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-f16.gguf) | f16 | 29.55GB | false | Full F16 weights. |
-| [SuperNova-14B-Q8_0.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q8_0.gguf) | Q8_0 | 15.70GB | false | Extremely high quality, generally unneeded but max available quant. |
-| [SuperNova-14B-Q6_K_L.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q6_K_L.gguf) | Q6_K_L | 12.50GB | false | Uses Q8_0 for embed and output weights. Very high quality, near perfect, *recommended*. |
-| [SuperNova-14B-Q6_K.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q6_K.gguf) | Q6_K | 12.12GB | false | Very high quality, near perfect, *recommended*. |
-| [SuperNova-14B-Q5_K_L.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q5_K_L.gguf) | Q5_K_L | 10.99GB | false | Uses Q8_0 for embed and output weights. High quality, *recommended*. |
-| [SuperNova-14B-Q5_K_M.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q5_K_M.gguf) | Q5_K_M | 10.51GB | false | High quality, *recommended*. |
-| [SuperNova-14B-Q5_K_S.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q5_K_S.gguf) | Q5_K_S | 10.27GB | false | High quality, *recommended*. |
-| [SuperNova-14B-Q4_K_L.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q4_K_L.gguf) | Q4_K_L | 9.57GB | false | Uses Q8_0 for embed and output weights. Good quality, *recommended*. |
-| [SuperNova-14B-Q4_K_M.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q4_K_M.gguf) | Q4_K_M | 8.99GB | false | Good quality, default size for must use cases, *recommended*. |
-| [SuperNova-14B-Q3_K_XL.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q3_K_XL.gguf) | Q3_K_XL | 8.61GB | false | Uses Q8_0 for embed and output weights. Lower quality but usable, good for low RAM availability. |
-| [SuperNova-14B-Q4_K_S.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q4_K_S.gguf) | Q4_K_S | 8.57GB | false | Slightly lower quality with more space savings, *recommended*. |
-| [SuperNova-14B-Q4_0.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q4_0.gguf) | Q4_0 | 8.54GB | false | Legacy format, generally not worth using over similarly sized formats |
-| [SuperNova-14B-Q4_0_8_8.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q4_0_8_8.gguf) | Q4_0_8_8 | 8.52GB | false | Optimized for ARM inference. Requires 'sve' support (see link below). *Don't use on Mac or Windows*. |
-| [SuperNova-14B-Q4_0_4_8.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q4_0_4_8.gguf) | Q4_0_4_8 | 8.52GB | false | Optimized for ARM inference. Requires 'i8mm' support (see link below). *Don't use on Mac or Windows*. |
-| [SuperNova-14B-Q4_0_4_4.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q4_0_4_4.gguf) | Q4_0_4_4 | 8.52GB | 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*. |
-| [SuperNova-14B-IQ4_XS.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-IQ4_XS.gguf) | IQ4_XS | 8.12GB | false | Decent quality, smaller than Q4_K_S with similar performance, *recommended*. |
-| [SuperNova-14B-Q3_K_L.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q3_K_L.gguf) | Q3_K_L | 7.92GB | false | Lower quality but usable, good for low RAM availability. |
-| [SuperNova-14B-Q3_K_M.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q3_K_M.gguf) | Q3_K_M | 7.34GB | false | Low quality. |
-| [SuperNova-14B-IQ3_M.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-IQ3_M.gguf) | IQ3_M | 6.92GB | false | Medium-low quality, new method with decent performance comparable to Q3_K_M. |
-| [SuperNova-14B-Q3_K_S.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q3_K_S.gguf) | Q3_K_S | 6.66GB | false | Low quality, not recommended. |
-| [SuperNova-14B-Q2_K_L.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q2_K_L.gguf) | Q2_K_L | 6.53GB | false | Uses Q8_0 for embed and output weights. Very low quality but surprisingly usable. |
-| [SuperNova-14B-IQ3_XS.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-IQ3_XS.gguf) | IQ3_XS | 6.38GB | false | Lower quality, new method with decent performance, slightly better than Q3_K_S. |
-| [SuperNova-14B-Q2_K.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-Q2_K.gguf) | Q2_K | 5.77GB | false | Very low quality but surprisingly usable. |
-| [SuperNova-14B-IQ2_M.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-IQ2_M.gguf) | IQ2_M | 5.36GB | false | Relatively low quality, uses SOTA techniques to be surprisingly usable. |
-| [SuperNova-14B-IQ2_S.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-IQ2_S.gguf) | IQ2_S | 5.00GB | false | Low quality, uses SOTA techniques to be usable. |
-| [SuperNova-14B-IQ2_XS.gguf](https://huggingface.co/bartowski/SuperNova-14B-GGUF/blob/main/SuperNova-14B-IQ2_XS.gguf) | IQ2_XS | 4.70GB | 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/SuperNova-14B-GGUF --include "SuperNova-14B-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/SuperNova-14B-GGUF --include "SuperNova-14B-Q8_0/*" --local-dir ./
-```
-You can either specify a new local-dir (SuperNova-14B-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

 ---
+base_model:
+- Qwen/Qwen2.5-14B
+library_name: transformers
 tags:
 - mergekit
 - merge
+license: apache-2.0
 ---
+# Arcee-SuperNova-Medius
+Arcee-SuperNova-Medius is a 14B parameter language model developed by Arcee.ai, built on the Qwen2.5-14B-Instruct architecture. This unique model is the result of a cross-architecture distillation pipeline, combining knowledge from both the Qwen2.5-72B-Instruct model and the Llama-3.1-405B-Instruct model. By leveraging the strengths of these two distinct architectures, SuperNova-Medius achieves high-quality instruction-following and complex reasoning capabilities in a mid-sized, resource-efficient form.
+SuperNova-Medius is designed to excel in a variety of business use cases, including customer support, content creation, and technical assistance, while maintaining compatibility with smaller hardware configurations. It’s an ideal solution for organizations looking for advanced capabilities without the high resource requirements of larger models like our SuperNova-70B.
+## Distillation Overview
+The development of SuperNova-Medius involved a sophisticated multi-teacher, cross-architecture distillation process, with the following key steps:
+1. **Logit Distillation from Llama-3.1-405B-Instruct**:
+   - We distilled the logits of Llama-3.1-405B-Instruct to Qwen2.5-14B using KL-divergence as the loss function. This allowed us to capture the nuanced distribution of Llama's outputs while adapting them to Qwen's architecture.
+2. **Logit and Hidden State Distillation from Qwen2.5-72B-Instruct**:
+   - Further distillation was performed using a combination of logit and hidden state distillation from Qwen2.5-72B-Instruct to ensure that SuperNova-Medius inherited the strong instruction-following capabilities and domain-specific knowledge of Qwen2.5.
+3. **Cross-Architecture Vocabulary Alignment**:
+   - Using `mergekit-tokensurgeon`, we aligned the vocabularies and hidden states of both teacher models, allowing for seamless integration of knowledge across the different architectures. This enabled SuperNova-Medius to effectively combine the strengths of both models.
+4. **Final Fusion and Fine-Tuning**:
+   - After aligning the vocabularies, a final fusion and fine-tuning step was conducted, using a specialized dataset from [EvolKit](https://github.com/arcee-ai/EvolKit) to ensure that SuperNova-Medius maintained coherence, fluency, and context understanding across a broad range of tasks.
+## Performance Evaluation
+Below are the benchmark results of SuperNova-Medius compared to similar models in its class:
+| Model | Average | IFEval | BBH | GPQA | MMLU Pro | MuSR | Math Level 5 |
+| --- | --- | --- | --- | --- | --- | --- | --- |
+| Mistral-Small 2409 | 0.423 | 0.628 | 0.581 | 0.333 | 0.410 | 0.406 | 0.181 |
+| Supernova-Lite | 0.427 | 0.786 | 0.511 | 0.306 | 0.388 | 0.415 | 0.155 |
+| Qwen2.5-14B-Instruct | 0.450 | 0.827 | 0.623 | 0.358 | 0.490 | 0.403 | 0.000 |
+| Supernova-Medius | **0.480** | **0.832** | **0.631** | **0.359** | **0.502** | **0.402** | **0.152** |
+SuperNova-Medius performs exceptionally well in instruction-following (IFEval) and complex reasoning tasks (BBH), demonstrating its capability to handle a variety of real-world scenarios. It outperforms Qwen2.5-14B and SuperNova-Lite in multiple benchmarks, making it a powerful yet efficient choice for high-quality generative AI applications.
+## Model Use Cases
+Arcee-SuperNova-Medius is suitable for a range of applications, including:
+- **Customer Support**: With its robust instruction-following and dialogue management capabilities, SuperNova-Medius can handle complex customer interactions, reducing the need for human intervention.
+- **Content Creation**: The model’s advanced language understanding and generation abilities make it ideal for creating high-quality, coherent content across diverse domains.
+- **Technical Assistance**: SuperNova-Medius has a deep reservoir of technical knowledge, making it an excellent assistant for programming, technical documentation, and other expert-level content creation.
+## Deployment Options
+SuperNova-Medius is available for use under the Apache-2.0 license. For those who need even higher performance, the full-size 70B SuperNova model can be accessed via an Arcee-hosted API or for local deployment. To learn more or explore deployment options, please reach out to [[email protected]](mailto:[email protected]).
+## Technical Specifications
+- **Model Architecture**: Qwen2.5-14B-Instruct
+- **Distillation Sources**: Qwen2.5-72B-Instruct, Llama-3.1-405B-Instruct
+- **Parameter Count**: 14 billion
+- **Training Dataset**: Custom instruction dataset generated with [EvolKit](https://github.com/arcee-ai/EvolKit)
+- **Distillation Technique**: Multi-architecture logit and hidden state distillation with cross-architecture vocabulary alignment.
+## Summary
+Arcee-SuperNova-Medius provides a unique balance of power, efficiency, and versatility. By distilling knowledge from two top-performing teacher models into a single 14B parameter model, SuperNova-Medius achieves results that rival larger models while maintaining a compact size ideal for practical deployment. Whether for customer support, content creation, or technical assistance, SuperNova-Medius is the perfect choice for organizations looking to leverage advanced language model capabilities in a cost-effective and accessible form.