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

Kancil V1 Llama3 4bit by afrizalha

 »  All LLMs  »  afrizalha  »  Kancil V1 Llama3 4bit   URL Share it on

  89   
  4-bit   4bit   Autotrain compatible   Bitsandbytes Dataset:catinthebag/tumpeng-1-...   Id   Indonesia   Llama   Llama3   Quantized   Region:us   Safetensors   Sharded   Tensorflow   Unsloth
Model Card on HF 🤗: https://huggingface.co/afrizalha/Kancil-V1-llama3-4bit 

Kancil V1 Llama3 4bit Benchmarks

LLME Score: 0.19264
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Kancil V1 Llama3 4bit (afrizalha/Kancil-V1-llama3-4bit)

Kancil V1 Llama3 4bit Parameters and Internals

Model Type 
Generative, Instruction-tuned
Use Cases 
Areas:
Research, AI Hobbyists
Primary Use Cases:
Multi-turn conversation, Text generation, QA from text, Making slogans
Limitations:
Minimal safety curation
Considerations:
This model is for research preview and fun; not for commercial use.
Additional Notes 
An issue with the dataset where newline characters are incorrectly interpreted.
Supported Languages 
id (proficient)
Training Details 
Data Sources:
catinthebag/Tumpeng-1-Indonesian
Data Volume:
14.8 million words
Methodology:
Fine-tuned with QLoRA using Unsloth framework
Context Length:
4096
Hardware Used:
GPU (at least T4)
Model Architecture:
Llama 3 8B with adapter merged
Input Output 
Input Format:
<|user|> {prompt} <|assistant|> {response}
Output Format:
Generated text
Performance Tips:
Keep the .replace() method in code to fix newline character errors.
Release Notes 
Version:
1.0
Notes:
Second working prototype with new features like multi-turn conversation and better text generation.
LLM NameKancil V1 Llama3 4bit
Repository 🤗https://huggingface.co/afrizalha/Kancil-V1-llama3-4bit 
Model Size8b
Required VRAM5.8 GB
Updated2025-02-22
Maintainerafrizalha
Model Typellama
Model Files  4.7 GB: 1-of-2   1.1 GB: 2-of-2
Supported Languagesid
Quantization Type4bit
Model ArchitectureLlamaForCausalLM
Licensellama3
Context Length8192
Model Max Length8192
Transformers Version4.41.1
Tokenizer ClassPreTrainedTokenizerFast
Padding Token<|reserved_special_token_250|>
Vocabulary Size128256
Torch Data Typebfloat16

Best Alternatives to Kancil V1 Llama3 4bit

Best Alternatives
Context / RAM
Downloads
Likes
...B Instruct Gradient 1048K 4bit1024K / 4.5 GB212
...B Instruct Gradient 1048K 8bit1024K / 8.6 GB71
...truct Gradient 1048K Bpw6 EXL21024K / 6.7 GB102
...truct Gradient 1048K Bpw5 EXL21024K / 5.8 GB70
Llama 3 8B Instruct 1048K 4bit1024K / 4.5 GB1225
Llama 3 8B Instruct 1048K 8bit1024K / 8.6 GB2817
... Gradient 1048K 8.0bpw H8 EXL21024K / 8.6 GB83
...ct Gradient 1048K Bpw2.25 EXL21024K / 3.4 GB51
Llama 3 8B Instruct 262K 2bit256K / 2.5 GB71
...B Instruct 262k V2 EXL2 6.0bpw256K / 6.7 GB111
Note: green Score (e.g. "73.2") means that the model is better than afrizalha/Kancil-V1-llama3-4bit.

Rank the Kancil V1 Llama3 4bit 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