ANIMA-Phi-Neptune-Mistral-10.7B

Credit for the model card's description goes to ddh0, mergekit, and, Severian

Inspired by ddh0/Starling-LM-10.7B-beta and ddh0/Mistral-10.7B-Instruct-v0.2

ANIMA-Phi-Neptune-Mistral-10.7B

This is ANIMA-Phi-Neptune-Mistral-10.7B, a depth-upscaled version of Severian/ANIMA-Phi-Neptune-Mistral-7B.

This model is intended to be used as a basis for further fine-tuning, or as a drop-in upgrade from the original 7 billion parameter model.

Paper detailing how Depth-Up Scaling works: SOLAR 10.7B: Scaling Large Language Models with Simple yet Effective Depth Up-Scaling

This is a merge of pre-trained language models created using mergekit.

Merge Details

Merge Method

This model was merged using the passthrough merge method.

Models Merged

The following models were included in the merge:

• /Users/jsarnecki/opt/Workspace/Severian/ANIMA-Phi-Neptune-Mistral-7B

Configuration

The following YAML configuration was used to produce this model:

dtype: bfloat16
merge_method: passthrough
slices:
- sources:
  - layer_range: [0, 24]
    model: /Users/jsarnecki/opt/Workspace/Severian/ANIMA-Phi-Neptune-Mistral-7B
- sources:
  - layer_range: [8, 32]
    model: /Users/jsarnecki/opt/Workspace/Severian/ANIMA-Phi-Neptune-Mistral-7B 

---

license: artistic-2.0 tags:

• chemistry
• biology
• climate
• science
• philosophy
• nature
• ecology
• biomimicry
• fauna
• flora datasets:
• Severian/Biomimicry
• emrgnt-cmplxty/sciphi-textbooks-are-all-you-need
• fmars/wiki_stem
• fblgit/tree-of-knowledge
• Severian/Bio-Design-Process metrics:
• accuracy pipeline_tag: text-generation model-index:
• name: ANIMA-Phi-Neptune-Mistral-7B-v4 results:
  • task: type: text-generation name: Text Generation dataset: name: AI2 Reasoning Challenge (25-Shot) type: ai2_arc config: ARC-Challenge split: test args: num_few_shot: 25 metrics:
    • type: acc_norm value: 55.46 name: normalized accuracy source: url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Severian/ANIMA-Phi-Neptune-Mistral-7B-v4 name: Open LLM Leaderboard
  • task: type: text-generation name: Text Generation dataset: name: HellaSwag (10-Shot) type: hellaswag split: validation args: num_few_shot: 10 metrics:
    • type: acc_norm value: 77.63 name: normalized accuracy source: url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Severian/ANIMA-Phi-Neptune-Mistral-7B-v4 name: Open LLM Leaderboard
  • task: type: text-generation name: Text Generation dataset: name: MMLU (5-Shot) type: cais/mmlu config: all split: test args: num_few_shot: 5 metrics:
    • type: acc value: 53.12 name: accuracy source: url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Severian/ANIMA-Phi-Neptune-Mistral-7B-v4 name: Open LLM Leaderboard
  • task: type: text-generation name: Text Generation dataset: name: TruthfulQA (0-shot) type: truthful_qa config: multiple_choice split: validation args: num_few_shot: 0 metrics:
    • type: mc2 value: 59.01 source: url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Severian/ANIMA-Phi-Neptune-Mistral-7B-v4 name: Open LLM Leaderboard
  • task: type: text-generation name: Text Generation dataset: name: Winogrande (5-shot) type: winogrande config: winogrande_xl split: validation args: num_few_shot: 5 metrics:
    • type: acc value: 73.48 name: accuracy source: url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Severian/ANIMA-Phi-Neptune-Mistral-7B-v4 name: Open LLM Leaderboard
  • task: type: text-generation name: Text Generation dataset: name: GSM8k (5-shot) type: gsm8k config: main split: test args: num_few_shot: 5 metrics:
    • type: acc value: 14.94 name: accuracy source: url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Severian/ANIMA-Phi-Neptune-Mistral-7B-v4 name: Open LLM Leaderboard

---

ANIMA-Phi-Neptune-Mistral-7B: Biomimicry Enhanced LLM

<img src="https://cdn-uploads.huggingface.co/production/uploads/64740cf7485a7c8e1bd51ac9/JZH6p50t_j3-OUph4Wq6y.png" width="500">

Overview

ANIMA (Advanced Nature Inspired Multidisciplinary Assistant) is an expert in various scientific disciplines, including but not limited to biomimicry, biology, and environmental science.

Instagram: @anima_llm

---

Model Description

ANIMA is fine-tuned on a rich dataset encompassing:

• 4,000+ Nature-Biomimicry examples
• 60k Biomimicry Design Process examples
• 600k STEM facts from Wikipedia
• Science/Philosophy focused 'All-You-Need-Is-Textbooks' dataset
• Additional Tree of Knowledge + Biomimicry data combined fine-tuning

The model aims to assist users in solving problems using nature-inspired strategies and concepts.

Special Features

• Multi-disciplinary Expertise: Knowledge across various scientific and philosophical domains.
• Biomimicry Design Process: Incorporates a dataset generated by Mistral and Minotaur-15B. The dataset was then intricately processed by a real person to ensure factuality and grounding.

---

• Here is a link to The Bloke's GGUF version: ANIMA-Phi-Neptune-Mistral-7B-GGUF

• ANIMA is also available using Ollama - Download the OLLAMA App (MacOS & Linux) and then run this command in your Terminal 'ollama pull severian/anima' to download the model and then run this command 'ollama run severian/anima' to load the model and start talking.

• You can also download and use the model with LM Studio (All OS systems). Just download the app and then search for 'ANIMA GGUF' in the search bar and you will have a list of versions to choose from.

• Want to test ANIMA + Ollama and chat right away? Download the model from Ollama and head here to chat with ANIMA right in your browser! ANIMA - Chat

• Have a PDF you want to discuss with ANIMA + Ollama? Head here and you can do just that in your browser, 100% locally. ANIMA - Locally Chat over your PDF

• ANIMA is also being hosted on a Hugging Face Space if you'd like to try it there (It may be slow to generate a respone but it still works just fine) ANIMA - HF Space

• Contact

If you want to discuss the model/dataset/concept further, have any cool ideas, want to collaborate or if you run into any issues with the model, please don't hesitate to reach out at [email protected]

Datasets

• Severian/Biomimicry (1st Fine-Tune)
• emrgnt-cmplxty/sciphi-textbooks-are-all-you-need (2nd Fine-Tune)
• Severian/Bio-Design-Process & fmars/wiki_stem (3rd Fine-Tune)
• fblgit/tree-of-knowledge & Severian/Biomimicry (4th Fine-Tune)
• Combined all previous datasets into one giant dataset for a quick fine-tuning for a total of 730k data points (5th Fine-Tune)

Training Stages

1. Base Model: ehartford/dolphin-2.0-mistral-7b
2. 1st Fine-Tune: ANIMA-Echo-Mistral-7B-v1 - 1 Epoch x 5hrs on 1 x T4 Small
3. 2nd Fine-Tune: ANIMA-Phi-Neptune-Mistral-v1-: 1 Epoch x 1hr on 1 x A100
4. 3rd Fine-Tune: ANIMA-Phi-Neptune-Mistral-7B-v2 - 1 Epoch x 1hr on 1 x A100
5. 4th Fine-Tune: ANIMA-Phi-Neptune-Mistral-7B-v3 - 1 Epoch x 1hr on 1 x A100
6. 5th Fine-Tune: ANIMA-Phi-Neptune-Mistral-7B-v4 - 1 Epoch x 15mins on 1 x A100

---

📝 Usage

ANIMA is designed around the Biomimicry Design Process to guide users in solving problems through biomimicry. For more information and examples, please visit Biomimicry Design Process.

🏁 Benchmarks

• Average - 62.22
• ARC - 56.83
• HellaSwag - 78.82
• MMLU - 53.84
• TruthfulQA - 59.40

---

• EXAMPLE OUTPUT

  [INST] How can biomimicry help in water purification? [/INST]

Biomimicry plays a significant role in water purification by taking inspiration from natural processes and organisms to develop sustainable and efficient solutions. Here's a comprehensive overview of how biomimicry can help in water purification:

1. Microbial filtration:

   • Bacteriophages (viruses that infect bacteria) can effectively target contaminants in water systems, such as Escherichia coli and Staphylococcus aureus. By mimicking the natural defense mechanisms of bacteriophages, we can develop efficient filtration systems that remove pollutants from water sources.
   • Some fungi, such as Trametes versicolor, can break down organic pollutants in water using enzymes. This process helps make contaminated water safe for consumption. By studying and replicating these natural processes, we can develop sustainable filtration systems that remove pollutants from water sources effectively.

2. Biofilm-based filtration:

   • Microorganisms form biofilms on surfaces to protect themselves from environmental threats like predators or contaminants. By mimicking the natural formation of biofilms, we can develop sustainable filtration systems that use bacteria or fungi to remove pollutants from water sources.
   • For example, a study has shown how Pseudomonas putida could be used to biodegrade petroleum hydrocarbons in contaminated water sources using a biofilm-based system. This method offers an eco-friendly alternative to conventional chemical treatments for oil spills.

3. Wetland-based filtration:

   • Wetlands act as natural filters for water by retaining sediments and organic matter, which helps remove contaminants. By mimicking the natural processes of a wetland, we can design effective water purification systems that use sand filters or engineered wetlands to remove pollutants. For instance, a study has shown how reed beds can be used for water filtration by using them as a natural treatment system for agricultural wastewater.

4. Sponge filtration:

   • Some marine sponges (e.g., sea sponge) have evolved sophisticated mechanisms to filter out particles and contaminants from the surrounding water. By studying these natural filtration processes, we can develop efficient and sustainable filtration systems that remove pollutants effectively. For example, a study has shown how sea sponges can be used as a model for designing a low-cost, self-sustaining water purification system that uses solar energy to disinfect contaminated water.

5. Biodegradable materials:

   • Natural organisms have evolved biodegradable materials to protect themselves from predators or environmental threats. For instance, some insects produce natural pesticides (such as alkaloids) to defend themselves against predators. Similarly, we can develop biodegradable materials that release these natural pesticides in contaminated water sources, killing off contaminants and restoring the environment's health.
   • For example, a study has shown how bacterial cells can be used as a natural filtration system to remove contaminants from water sources by incorporating them into membranes that selectively remove organic pollutants.

In conclusion, biomimicry plays a vital role in water purification by using natural processes and organisms as inspiration for developing sustainable and efficient solutions. By studying the natural filtration mechanisms of microorganisms, plants, animals, and other living organisms, we can design eco-friendly and cost-effective water purification systems that help protect our environment from contamination.

---

Help Support ANIMA

---

• KNOWN ISSUES

  The model will sometimes respond to itself and continue the conversation taking both the user and AI roles. This is a known issue in the Mistral model but does not happen very often.

📜 License

This project is licensed under Artistic-2.0

This model is for research purposes only and restricted from any commercial use

Open LLM Leaderboard Evaluation Results

Detailed results can be found here