Solar Ko Recovery 11B by beomi

 ยป  All LLMs  ยป  beomi  ยป  Solar Ko Recovery 11B   URL Share it on

  Autotrain compatible Base model:finetune:upstage/so... Base model:upstage/solar-10.7b...   En   Ko   Llama   Mistral   Pytorch   Region:us   Safetensors   Sharded   Solar   Solar-ko   Tensorflow

Solar Ko Recovery 11B Benchmarks

nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Solar Ko Recovery 11B (beomi/Solar-Ko-Recovery-11B)

Solar Ko Recovery 11B Parameters and Internals

Model Type 
text generation
Use Cases 
Areas:
research, commercial applications
Additional Notes 
Vocab expansion conducted using edited upstage/solar-1-mini-tokenizer with enhancements in Korean and Japanese vocab.
Supported Languages 
Korean (advanced), English (intermediate)
Training Details 
Data Sources:
A curated mix of Korean+English Corpora
Data Volume:
>100B tokens
Methodology:
2-step training process: Only Embedding layer and LM Head layer are trained, then full parameters trained
Context Length:
4000
Model Architecture:
auto-regressive language model leveraging an optimized transformer architecture derived from Llama-2
Input Output 
Input Format:
text
Accepted Modalities:
text
Output Format:
text
Release Notes 
Version:
2024.07.01
Date:
2024-07-01
Notes:
Released Solar-Ko-Recovery & Uploaded Benchmark scores
Version:
2024.05.16
Date:
2024-05-16
Notes:
Preview Released Solar-Ko-Recovery
LLM NameSolar Ko Recovery 11B
Repository ๐Ÿค—https://huggingface.co/beomi/Solar-Ko-Recovery-11B 
Base Model(s)  upstage/SOLAR-10.7B-v1.0   upstage/SOLAR-10.7B-v1.0
Model Size11b
Required VRAM21.9 GB
Updated2025-02-05
Maintainerbeomi
Model Typellama
Model Files  2.9 GB: 1-of-8   2.9 GB: 2-of-8   3.0 GB: 3-of-8   2.9 GB: 4-of-8   2.9 GB: 5-of-8   2.9 GB: 6-of-8   3.0 GB: 7-of-8   1.4 GB: 8-of-8
Supported Languagesko en
Model ArchitectureLlamaForCausalLM
Licenseapache-2.0
Context Length4096
Model Max Length4096
Transformers Version4.33.1
Tokenizer ClassLlamaTokenizer
Vocabulary Size64000
Torch Data Typebfloat16

Best Alternatives to Solar Ko Recovery 11B

Best Alternatives
Context / RAM
Downloads
Likes
...ral 11B Omni OP 1K 2048 Ver0.132K / 21.4 GB510
MIstral 11B Omni OP U1k Ver0.132K / 21.4 GB470
Llama 3 Synatra 11B V1 20K20K / 23 GB79
Fimbulvetr 11B V2.1 16K16K / 21.4 GB4617
Moistral 11B V28K / 21.4 GB5921
Narumashi 11B V0.98K / 21.4 GB541
Moistral 11B V38K / 21.4 GB26195
Moistral 11B V5d E48K / 21.4 GB151
Moistral 11B V5a8K / 21.4 GB201
Moistral 11B V5b8K / 21.4 GB121
Note: green Score (e.g. "73.2") means that the model is better than beomi/Solar-Ko-Recovery-11B.

Rank the Solar Ko Recovery 11B 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