LLM EXPLORER
 LLM Explorer: A Curated Large Language Model Directory and Analytics
Dark Theme
  LLM  List    LLM Hosting    LLM Leaderboards    Blog    Newsfeed    Advertise

MAmmoTH2 8x7B Plus by TIGER-Lab

 »  All LLMs  »  TIGER-Lab  »  MAmmoTH2 8x7B Plus   URL Share it on

  3640   
  Merged Model   Arxiv:2405.03548   Autotrain compatible   Conversational Dataset:tiger-lab/webinstructs...   En   Endpoints compatible   Mixtral   Moe   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/TIGER-Lab/MAmmoTH2-8x7B-Plus 

MAmmoTH2 8x7B Plus Benchmarks

LLME Score: 0.22039
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
MAmmoTH2 8x7B Plus (TIGER-Lab/MAmmoTH2-8x7B-Plus)

MAmmoTH2 8x7B Plus Parameters and Internals

Model Type 
text generation, instruction tuning
Use Cases 
Areas:
research, commercial applications
Applications:
reasoning benchmarks, chatbot benchmarks
Primary Use Cases:
enhancing reasoning abilities of LLMs, instruction tuning
Limitations:
Performance may vary based on the complexity and specifics of the math problem., Not all mathematical fields are covered comprehensively.
Additional Notes 
Models show improvement in performance after fine-tuning on WEBINSTRUCT.
Supported Languages 
en (English - High proficiency)
Training Details 
Data Sources:
https://huggingface.co/datasets/TIGER-Lab/WebInstructSub
Methodology:
Fine-tuning with the WEBINSTRUCT dataset.
Input Output 
Input Format:
Prompting with instructions formatted using special tokens `~~[INST]` and `[/INST]`.
Accepted Modalities:
text
Output Format:
Generated text responses
Performance Tips:
Model is not very sensitive to the chat template.
Release Notes 
Version:
MAmmoTH2-7B
Notes:
Achieves performance of 36.7% on MATH and 68.4% on GSM8K.
Version:
MAmmoTH2-8B
Notes:
Achieves performance of 35.8% on MATH and 70.4% on GSM8K.
Version:
MAmmoTH2-8x7B
Notes:
Achieves outstanding performance with a variety of datasets.
LLM NameMAmmoTH2 8x7B Plus
Repository 🤗https://huggingface.co/TIGER-Lab/MAmmoTH2-8x7B-Plus 
Merged ModelYes
Model Size46.7b
Required VRAM93.6 GB
Updated2025-02-22
MaintainerTIGER-Lab
Model Typemixtral
Model Files  4.9 GB: 1-of-19   5.0 GB: 2-of-19   5.0 GB: 3-of-19   4.9 GB: 4-of-19   5.0 GB: 5-of-19   5.0 GB: 6-of-19   4.9 GB: 7-of-19   5.0 GB: 8-of-19   5.0 GB: 9-of-19   4.9 GB: 10-of-19   5.0 GB: 11-of-19   5.0 GB: 12-of-19   5.0 GB: 13-of-19   4.9 GB: 14-of-19   5.0 GB: 15-of-19   5.0 GB: 16-of-19   4.9 GB: 17-of-19   5.0 GB: 18-of-19   4.2 GB: 19-of-19   0.0 GB
Supported Languagesen
Model ArchitectureMixtralForCausalLM
Licensemit
Context Length32768
Model Max Length32768
Transformers Version4.40.0
Tokenizer ClassLlamaTokenizer
Padding Token</s>
Vocabulary Size32000
Torch Data Typebfloat16

Best Alternatives to MAmmoTH2 8x7B Plus

Best Alternatives
Context / RAM
Downloads
Likes
Mixtral 8x7B Instruct V0.132K / 93.6 GB5293684311
Nous Hermes 2 Mixtral 8x7B DPO32K / 93.6 GB2915424
Mixtral 8x7B V0.132K / 93.6 GB308741677
GritLM 8x7B KTO32K / 93.6 GB36673
Smaug Mixtral V0.132K / 187.7 GB362712
Sensualize Mixtral Bf1632K / 93.6 GB00
Skadi Mixtral V132K / 93.5 GB00
Franziska Mixtral V132K / 93.5 GB00
Typhon Mixtral V132K / 93.4 GB00
Merge Mixtral Prometheus 8x7B32K / 91.9 GB122
Note: green Score (e.g. "73.2") means that the model is better than TIGER-Lab/MAmmoTH2-8x7B-Plus.

Rank the MAmmoTH2 8x7B Plus Capabilities

🆘 Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! 🌟

Instruction Following and Task Automation  
Factuality and Completeness of Knowledge  
Censorship and Alignment  
Data Analysis and Insight Generation  
Text Generation  
Text Summarization and Feature Extraction  
Code Generation  
Multi-Language Support and Translation  

What open-source LLMs or SLMs are you in search of? 43470 in total.

Email us: info@extractum.io. Our Privacy Policy  | Terms and Conditions  | Suggest an improvement.
Our Social Media →  
Original data from HuggingFace, OpenCompass and various public git repos.
Release v20241227
  Support LLM Explorer