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

L3 8B Stheno V3.3 32K 8.0bpw H8 EXL2 by Pyroserenus

 »  All LLMs  »  Pyroserenus  »  L3 8B Stheno V3.3 32K 8.0bpw H8 EXL2   URL Share it on

  8   
  8-bit   Autotrain compatible   Conversational   En   Endpoints compatible   Exl2   Llama   Quantized   Region:us   Safetensors
Model Card on HF 🤗: https://huggingface.co/Pyroserenus/L3-8B-Stheno-v3.3-32K-8.0bpw-h8-exl2 

L3 8B Stheno V3.3 32K 8.0bpw H8 EXL2 Benchmarks

LLME Score: 0.16204
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
L3 8B Stheno V3.3 32K 8.0bpw H8 EXL2 (Pyroserenus/L3-8B-Stheno-v3.3-32K-8.0bpw-h8-exl2)
🌟 Advertise your project 🚀

L3 8B Stheno V3.3 32K 8.0bpw H8 EXL2 Parameters and Internals

Model Type 
text-generation
Use Cases 
Primary Use Cases:
Roleplaying, Creative Writing
Limitations:
Issues with long-context understanding and reasoning
Additional Notes 
Model use is facilitated by configurations inspired by winglian/Llama-3-8b-64k-PoSE. The model performs well despite its complexity compared to peers like RULER or NIAN.
Supported Languages 
en (fluent)
Training Details 
Data Sources:
Roleplaying Samples, Creative Writing Samples, Instruct Data
Methodology:
Trained at 8K Context -> Expanded to 32K Context with PoSE training. Dataset modifications included cleaning up roleplaying samples, removing low-quality samples, and adding more creative writing samples.
Context Length:
32768
Input Output 
Input Format:
Expected within configurations defined in the code snippet.
Accepted Modalities:
text
Output Format:
Text generated in response to the input.
Performance Tips:
Using configurations sourced from winglian's Llama-3-8b-64k-PoSE resulted in optimal performance metrics during training.
LLM NameL3 8B Stheno V3.3 32K 8.0bpw H8 EXL2
Repository 🤗https://huggingface.co/Pyroserenus/L3-8B-Stheno-v3.3-32K-8.0bpw-h8-exl2 
Model Size8b
Required VRAM8.6 GB
Updated2025-09-23
MaintainerPyroserenus
Model Typellama
Model Files  8.6 GB
Supported Languagesen
Quantization Typeexl2
Model ArchitectureLlamaForCausalLM
Licensecc-by-nc-4.0
Context Length32768
Model Max Length32768
Transformers Version4.41.1
Tokenizer ClassPreTrainedTokenizerFast
Padding Token<|end_of_text|>
Vocabulary Size128256
Torch Data Typebfloat16

Best Alternatives to L3 8B Stheno V3.3 32K 8.0bpw H8 EXL2

Best Alternatives
Context / RAM
Downloads
Likes
...B Instruct Gradient 1048K 4bit1024K / 4.5 GB62
...B Instruct Gradient 1048K 8bit1024K / 8.6 GB61
...truct Gradient 1048K Bpw6 EXL21024K / 6.7 GB102
...truct Gradient 1048K Bpw5 EXL21024K / 5.8 GB60
Llama 3 8B Instruct 1048K 4bit1024K / 4.5 GB17225
Llama 3 8B Instruct 1048K 8bit1024K / 8.6 GB10417
... Gradient 1048K 8.0bpw H8 EXL21024K / 8.6 GB103
...ct Gradient 1048K Bpw2.25 EXL21024K / 3.4 GB31
Llama 3 8B Instruct 262K 2bit256K / 2.5 GB71
...B Instruct 262k V2 EXL2 5.0bpw256K / 5.8 GB61
Note: green Score (e.g. "73.2") means that the model is better than Pyroserenus/L3-8B-Stheno-v3.3-32K-8.0bpw-h8-exl2.

Rank the L3 8B Stheno V3.3 32K 8.0bpw H8 EXL2 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? 51544 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 v20241124
  Support LLM Explorer