LLM EXPLORER 66,978 MODELS INDEXED

Llama 3.1 Swallow 8B Instruct V0.5 DFlash by florianvoss

By florianvoss · 40 downloads

Llama 3.1 Swallow 8B Instruct V0.5 DFlash is an open-source language model by florianvoss. Features: 8b LLM, VRAM: 2.1GB, License: other, Instruction-Based, LLM Explorer Score: 0.26.

  Arxiv:2602.06036 Base model:finetune:tokyotech-... Base model:tokyotech-llm/llama...   Bf16   Conversational   Custom code Dataset:keisukemiyamoto/lambda...   Dataset:llm-jp/magpie-sft-v1.0 Dataset:llm-jp/wizardlm8x22b-l... Dataset:tokyotech-llm/lmsys-ch...   Dflash   En   Instruct   Ja   Japanese   Region:us   Safetensors   Speculative-decoding   Speculators   Vllm

Llama 3.1 Swallow 8B Instruct V0.5 DFlash Parameters and Internals

LLM NameLlama 3.1 Swallow 8B Instruct V0.5 DFlash
Repository πŸ€—https://huggingface.co/florianvoss/Llama-3.1-Swallow-8B-Instruct-v0.5-DFlash 
Base Model(s)  tokyotech-llm/Llama-3.1-Swallow-8B-Instruct-v0.5   z-lab/LLaMA3.1-8B-Instruct-DFlash-UltraChat   tokyotech-llm/Llama-3.1-Swallow-8B-Instruct-v0.5   z-lab/LLaMA3.1-8B-Instruct-DFlash-UltraChat
Model Size8b
Required VRAM2.1 GB
Updated2026-10-22
Maintainerflorianvoss
Instruction-BasedYes
Model Files  2.1 GB
Supported Languagesja en
Model ArchitectureDFlashDraftModel
Licenseother
Transformers Version5.16.1
Tokenizer ClassPreTrainedTokenizerFast
Padding Token<|finetune_right_pad_id|>

Best Alternatives to Llama 3.1 Swallow 8B Instruct V0.5 DFlash

Best Alternatives
Context / RAM
Downloads
Likes
...1 8B Instruct DFlash UltraChat40K / 2.1 GB231764
...8B Instruct DFlash Safetensors40K / 1.1 GB2350
Note: green Score (e.g. "73.2") means that the model is better than florianvoss/Llama-3.1-Swallow-8B-Instruct-v0.5-DFlash.