LLM EXPLORER 59,516 MODELS INDEXED

SOLAR 10.7B Instruct V1.0 by upstage

By upstage · 53386 downloads
HF Repo ↗ Copy URL

SOLAR 10.7B Instruct V1.0 is an open-source language model by upstage. Features: 10.7b LLM, VRAM: 21.4GB, Context: 4K, License: cc-by-nc-4.0, Instruction-Based, LLM Explorer Score: 0.35, ELO: 1077, Arc: 71.1, HellaSwag: 88.2, MMLU: 66.2, GSM8K: 64.8.

  Arxiv:2309.12284   Arxiv:2312.15166   Arxiv:2403.19270 Base model:finetune:upstage/so... Base model:upstage/solar-10.7b...   Conversational Dataset:allenai/ultrafeedback ... Dataset:c-s-ale/alpaca-gpt4-da...   Dataset:intel/orca dpo pairs   Dataset:open-orca/openorca   Deploy:azure   En   Endpoints compatible   Instruct   Llama   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/upstage/SOLAR-10.7B-Instruct-v1.0 

SOLAR 10.7B Instruct V1.0 Benchmarks

LMSys ELO 1077 vs 1272 (so35) -15.3%
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

SOLAR 10.7B Instruct V1.0 Parameters and Internals

Model Type 
large language model
Use Cases 
Areas:
research, conversational AI
Primary Use Cases:
single-turn conversation
Limitations:
Not suitable for multi-turn conversation
Supported Languages 
en (full support)
Training Details 
Data Sources:
c-s-ale/alpaca-gpt4-data, Open-Orca/OpenOrca, Intel/orca_dpo_pairs, allenai/ultrafeedback_binarized_cleaned
Methodology:
Supervised fine-tuning (SFT) and Direct Preference Optimization (DPO)
Model Architecture:
Integration of Mistral 7B weights for depth up-scaling
Input Output 
Input Format:
Single-turn conversation
Accepted Modalities:
text
Output Format:
text
LLM NameSOLAR 10.7B Instruct V1.0
Repository 🤗https://huggingface.co/upstage/SOLAR-10.7B-Instruct-v1.0 
Base Model(s)  upstage/SOLAR-10.7B-v1.0   upstage/SOLAR-10.7B-v1.0
Model Size10.7b
Required VRAM21.4 GB
Updated2026-07-29
Maintainerupstage
Model Typellama
Instruction-BasedYes
Model Files  4.9 GB: 1-of-5   5.0 GB: 2-of-5   4.9 GB: 3-of-5   4.9 GB: 4-of-5   1.7 GB: 5-of-5
Supported Languagesen
Model ArchitectureLlamaForCausalLM
Licensecc-by-nc-4.0
Context Length4096
Model Max Length4096
Transformers Version4.35.0
Tokenizer ClassLlamaTokenizer
Padding Token</s>
Vocabulary Size32000
Torch Data Typefloat16

Quantized Models of the SOLAR 10.7B Instruct V1.0

Model
Likes
Downloads
VRAM
SOLAR 10.7B Instruct V1.0 GGUF8415224 GB
UNA SOLAR 10.7B Instruct V1.01619721 GB
SOLAR 10.7B Instruct V1.0 GGUF25814 GB
SOLAR 10.7B Instruct V1.0 GPTQ246 GB
SOLAR 10.7B Instruct V1.0 AWQ51236 GB
SOLAR 10.7B Instruct V1.0 GPTQ12146 GB

Best Alternatives to SOLAR 10.7B Instruct V1.0

Best Alternatives
Context / RAM
Downloads
Likes
SOLAR 10.7B Instruct V1.0 128K128K / 21.4 GB573
SOLAR 10.7B V1.0 Instruct 16K16K / 21.4 GB632
SauerkrautLM SOLAR Instruct8K / 21.4 GB6246
MetaModelv28K / 21.4 GB1970
Kazemi 1.2 Solar8K / 21.4 GB00
Solarmer34K / 21.4 GB50
Somer24K / 21.4 GB100
SOLAR 10.7B Instruct V1.04K / 21.4 GB170
Somer4K / 21.4 GB80
...arbonVillain En 10.7B V2 Slerp4K / 21.4 GB81370
Note: green Score (e.g. "73.2") means that the model is better than upstage/SOLAR-10.7B-Instruct-v1.0.