Calm3 22B RP V0.1 by Aratako

 ยป  All LLMs  ยป  Aratako  ยป  Calm3 22B RP V0.1   URL Share it on

  Autotrain compatible Base model:cyberagent/calm3-22... Base model:finetune:cyberagent...   Conversational Dataset:aratako/rosebleu-1on1-... Dataset:aratako/synthetic-japa... Dataset:aratako/synthetic-japa... Dataset:chaser-cz/roleplay scr... Dataset:grimulkan/limarp-augme... Dataset:minervaai/aesir-previe...   Dataset:openerotica/freedom-rp   Dataset:openerotica/lima-nsfw Dataset:roleplay4fun/aesir-v1.... Dataset:sicariussicariistuff/b...   Endpoints compatible   Ja   Llama   Region:us   Safetensors   Sharded   Tensorflow   Trl   Unsloth

Calm3 22B RP V0.1 Benchmarks

nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Calm3 22B RP V0.1 (Aratako/calm3-22b-RP-v0.1)

Calm3 22B RP V0.1 Parameters and Internals

Model Type 
text-generation
Additional Notes 
Trained with Unsloth and Huggingface's TRL library for faster training.
Supported Languages 
ja (native or primary)
Training Details 
Data Sources:
Aratako/Rosebleu-1on1-Dialogues-RP, Aratako/Synthetic-Japanese-Roleplay-NSFW-Claude-3.5s-10.5k-formatted, Aratako/Synthetic-Japanese-Roleplay-gpt-4o-mini-19.8k-formatted, grimulkan/LimaRP-augmented, SicariusSicariiStuff/Bluemoon_Top50MB_Sorted_Fixed, MinervaAI/Aesir-Preview, openerotica/freedom-rp, openerotica/lima-nsfw, Chaser-cz/roleplay_scripts, roleplay4fun/aesir-v1.1
Methodology:
Fine-tuned using QLoRA on a single A10 GPU.
Context Length:
8192
Hardware Used:
A10 GPU
Input Output 
Input Format:
ChatML format; example provided.
Accepted Modalities:
text
Output Format:
Generated text based on roleplay input format
Performance Tips:
Use `tokenizer.apply_chat_template()` for proper input formatting.
LLM NameCalm3 22B RP V0.1
Repository ๐Ÿค—https://huggingface.co/Aratako/calm3-22b-RP-v0.1 
Base Model(s)  Calm3 22B Chat   cyberagent/calm3-22b-chat
Model Size22b
Required VRAM44.9 GB
Updated2025-01-28
MaintainerAratako
Model Typellama
Model Files  4.9 GB: 1-of-10   4.9 GB: 2-of-10   4.8 GB: 3-of-10   4.9 GB: 4-of-10   5.0 GB: 5-of-10   4.8 GB: 6-of-10   4.9 GB: 7-of-10   4.8 GB: 8-of-10   4.9 GB: 9-of-10   1.0 GB: 10-of-10
Supported Languagesja
Model ArchitectureLlamaForCausalLM
Licensecc-by-nc-sa-4.0
Context Length16384
Model Max Length16384
Transformers Version4.44.0
Tokenizer ClassGPTNeoXTokenizer
Padding Token<|padding|>
Vocabulary Size65024
Torch Data Typebfloat16

Best Alternatives to Calm3 22B RP V0.1

Best Alternatives
Context / RAM
Downloads
Likes
Calm3 22B Chat16K / 44.9 GB549872
Calm3 22B RP V216K / 44.9 GB13911
Yousei 22B4K / 44.5 GB13262
Llama2 22B Daydreamer V34K / 43.7 GB130011
Platypus 2 22B Relora4K / 43.7 GB12681
Llama2 22B4K / 43.7 GB130346
Llama2 22B Blocktriangular4K / 43.7 GB13664
Llama2 22B Daydreamer V24K / 43.7 GB162
Llama2 22B Daydreamer V14K / 43.7 GB102
Llama2 22B Empath Alpacagpt44K / 43.7 GB111
Note: green Score (e.g. "73.2") means that the model is better than Aratako/calm3-22b-RP-v0.1.

Rank the Calm3 22B RP V0.1 Capabilities

๐Ÿ†˜ 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! ๐ŸŒŸ

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? 42577 in total.

Our Social Media →  
Original data from HuggingFace, OpenCompass and various public git repos.
Release v20241227