miquliz-120b-v2.0
Maintainer: wolfram
85
👁️
| Property | Value |
|---|---|
| Run this model | Run on HuggingFace |
| API spec | View on HuggingFace |
| Github link | No Github link provided |
| Paper link | No paper link provided |
Create account to get full access
Model overview
The miquliz-120b-v2.0 is a 120 billion parameter large language model created by interleaving layers of the miqu-1-70b-sf and lzlv_70b_fp16_hf models using the mergekit tool. It was improved from the previous v1.0 version by incorporating techniques from the TheProfessor-155b model. The model is inspired by the goliath-120b and is maintained by Wolfram.
Model inputs and outputs
Inputs
- Text prompts of up to 32,768 tokens in length
Outputs
- Continuation of the provided text prompt, generating new relevant text
Capabilities
The miquliz-120b-v2.0 model is capable of impressive performance, achieving top ranks and double perfect scores in the maintainer's own language model comparisons and tests. It demonstrates strong general language understanding and generation abilities across a variety of tasks.
What can I use it for?
The large scale and high performance of the miquliz-120b-v2.0 model make it well-suited for language-related applications that require powerful text generation, such as content creation, question answering, and conversational AI. The model could be fine-tuned for specific domains or integrated into products via the CopilotKit open-source platform.
Things to try
Explore the model's capabilities by prompting it with a variety of tasks, from creative writing to analysis and problem solving. The model's size and breadth of knowledge make it an excellent starting point for developing custom language models tailored to your needs.
This summary was produced with help from an AI and may contain inaccuracies - check out the links to read the original source documents!
Related Models
⚙️
miqu-1-120b
48
The miqu-1-120b model is a 120B parameter language model created by Wolfram, the maintainer of the model. It is a "frankenmerge" model, meaning it was created by interleaving layers of the miqu-1-70b model, created by miqudev, with itself using the mergekit tool. The model was inspired by several other 120B models such as Venus-120b-v1.2, MegaDolphin-120b, and goliath-120b. Model inputs and outputs The miqu-1-120b model is a text-to-text transformer model, which means it can be used for a variety of natural language processing tasks such as generation, summarization, and translation. The model takes text prompts as input and generates relevant text as output. Inputs Text prompts of varying lengths, from a few words to multiple paragraphs Outputs Generated text in response to the input prompt, with lengths ranging from a few sentences to multiple paragraphs Capabilities The miqu-1-120b model is a large and powerful language model capable of producing coherent and context-appropriate text. It has demonstrated strong performance on a variety of benchmarks, including high scores on tasks like the AI2 Reasoning Challenge, HellaSwag, and Winogrande. What can I use it for? The miqu-1-120b model could be used for a wide range of natural language processing tasks, including: Creative writing**: The model's text generation capabilities make it well-suited for assisting with creative writing projects, such as short stories, poetry, and even collaborative worldbuilding. Conversational AI**: With its ability to engage in contextual and coherent dialogue, the model could be used to create more natural and engaging conversational AI assistants. Content generation**: The model could be employed to generate a variety of content, such as news articles, blog posts, or social media updates, with the potential for customization and personalization. Education and research**: Researchers and educators could use the model to explore natural language processing, test new techniques, or develop educational applications. Things to try One interesting aspect of the miqu-1-120b model is its ability to adapt to different prompting styles and templates. By experimenting with the Mistral prompt format, users can try to elicit different types of responses, from formal and informative to more creative and expressive. Additionally, the model's large size and high context capacity (up to 32,768 tokens) make it well-suited for longer-form tasks, such as generating detailed descriptions, worldbuilding, or interactive storytelling. Users could try providing the model with rich contextual information and see how it responds and builds upon the existing narrative.
Updated Invalid Date
🌀
Midnight-Miqu-70B-v1.5
75
The Midnight-Miqu-70B-v1.5 model is a DARE Linear merge between the sophosympatheia/Midnight-Miqu-70B-v1.0 and migtissera/Tess-70B-v1.6 models. This version is close in feel and performance to Midnight Miqu v1.0 but the maintainer believes it picked up some improvements from Tess. The model is uncensored, and the maintainer warns that users are responsible for how they use it. Model Inputs and Outputs Inputs Free-form text prompts of any length Outputs Continuation of the input prompt, generating coherent and contextually relevant text Capabilities The Midnight-Miqu-70B-v1.5 model is designed for roleplaying and storytelling, and the maintainer believes it performs well in these areas. It may also be capable of other text generation tasks, but the maintainer has not extensively tested its performance outside of creative applications. What Can I Use It For? The Midnight-Miqu-70B-v1.5 model could be useful for a variety of creative writing and roleplaying projects, such as writing interactive fiction, generating narrative content for games, or developing unique characters and stories. Its ability to produce long-form, contextually relevant text makes it well-suited for these types of applications. Things to Try One key capability of the Midnight-Miqu-70B-v1.5 model is its ability to handle long context windows, up to 32K tokens. Experimenting with different sampling techniques, such as Quadratic Sampling and Min-P, can help optimize the model's performance for creative use cases. Additionally, adjusting the repetition penalty and other parameters can lead to more diverse and engaging output.
Updated Invalid Date
🤔
Midnight-Miqu-70B-v1.0
48
The Midnight-Miqu-70B-v1.0 model is a merge between the 152334H/miqu-1-70b-sf and sophosympatheia/Midnight-Rose-70B-v2.0.3 models. It retains much of what made Midnight Rose special while gaining some long-context capabilities from Miqu. Model inputs and outputs The Midnight-Miqu-70B-v1.0 model is a text-to-text model, meaning it takes in text prompts and generates text outputs. It can handle long-form contexts up to 32,000 tokens. Inputs Text prompts of variable length Outputs Generated text continuations based on the input prompts Capabilities The Midnight-Miqu-70B-v1.0 model performs well at roleplaying and storytelling tasks. It can maintain coherence and authenticity in character actions, thoughts, and dialogue over long sequences. What can I use it for? The Midnight-Miqu-70B-v1.0 model is well-suited for creative writing and roleplaying applications. It could be used to collaboratively generate engaging fiction, worldbuild compelling narratives, or play out dynamic interactive stories. The model's long-context abilities make it valuable for tasks requiring sustained, cohesive output. Things to try You can experiment with the model's long-context capabilities by running it out to 32,000 tokens with an alpha_rope setting of 1. Limited testing shows it can maintain coherence even out to 64,000 tokens using an alpha_rope of 2.5. Additionally, try using Quadratic Sampling (smoothing factor) and Min-P sampling to optimize the model's creative output.
Updated Invalid Date
📈
Mixtral-8x7B-Instruct-v0.1-AWQ
54
The Mixtral-8x7B-Instruct-v0.1-AWQ is a language model created by Mistral AI_. It is an 8 billion parameter model that has been fine-tuned on instructional data, allowing it to follow complex prompts and generate relevant, coherent responses. Compared to similar large language models like Mixtral-8x7B-Instruct-v0.1-GPTQ and Mistral-7B-Instruct-v0.1-GPTQ, the Mixtral-8x7B-Instruct-v0.1-AWQ uses the efficient AWQ quantization method to provide faster inference with equivalent or better quality compared to common GPTQ settings. Model inputs and outputs The Mixtral-8x7B-Instruct-v0.1-AWQ is a text-to-text model, taking natural language prompts as input and generating relevant, coherent text as output. The model has been fine-tuned to follow specific instructions and prompts, allowing it to engage in tasks like open-ended storytelling, analysis, and task completion. Inputs Natural language prompts**: The model accepts free-form text prompts that can include instructions, queries, or open-ended requests. Instructional formatting**: The model responds best to prompts that use the [INST] and [/INST] tags to delineate the instructional component. Outputs Generated text**: The model's primary output is a continuation of the input prompt, generating relevant, coherent text that follows the given instructions or request. Contextual awareness**: The model maintains awareness of the broader context and can generate responses that build upon previous interactions. Capabilities The Mixtral-8x7B-Instruct-v0.1-AWQ model demonstrates strong capabilities in following complex prompts and generating relevant, coherent responses. It excels at open-ended tasks like storytelling, where it can continue a narrative in a natural and imaginative way. The model also performs well on analysis and task completion, providing thoughtful and helpful responses to a variety of prompts. What can I use it for? The Mixtral-8x7B-Instruct-v0.1-AWQ model can be a valuable tool for a wide range of applications, from creative writing and content generation to customer support and task automation. Its ability to understand and respond to natural language instructions makes it well-suited for chatbots, virtual assistants, and other interactive applications. One potential use case could be a creative writing assistant, where the model could help users brainstorm story ideas, develop characters, and expand upon plot points. Alternatively, the model could be used in a customer service context, providing personalized responses to inquiries and helping to streamline support workflows. Things to try Beyond the obvious use cases, there are many interesting things to explore with the Mixtral-8x7B-Instruct-v0.1-AWQ model. For example, you could try providing the model with more open-ended prompts to see how it responds, or challenge it with complex multi-step instructions to gauge its reasoning and problem-solving capabilities. Additionally, you could experiment with different sampling parameters, such as temperature and top-k, to find the settings that work best for your specific use case. Overall, the Mixtral-8x7B-Instruct-v0.1-AWQ is a powerful and versatile language model that can be a valuable tool in a wide range of applications. Its efficient quantization and strong performance on instructional tasks make it an attractive option for developers and researchers looking to push the boundaries of what's possible with large language models.
Updated Invalid Date