LLM EXPLORER 66,613 MODELS INDEXED

Qlora Self Instruct 65B by timdettmers

By timdettmers · 0 downloads
HF Repo ↗ Copy URL

Qlora Self Instruct 65B is an open-source language model by timdettmers. Features: 65b LLM, VRAM: 3.2GB, Instruction-Based, LLM Explorer Score: 0.07.

  Arxiv:2302.13971   Arxiv:2305.14314   Adapter   Finetuned   Instruct   Lora   Region:us
Model Card on HF 🤗: https://huggingface.co/timdettmers/qlora-self-instruct-65b 

Qlora Self Instruct 65B Parameters and Internals

Model Type 
causal language model
Use Cases 
Areas:
research
Additional Notes 
Models utilize NormalFloat4 datatype for base model and LoRA adapters on all linear layers with BFloat16 for computation.
Training Details 
Data Sources:
Alpaca, HH-RLHF, Unnatural Instr., Chip2, Longform, Self-Instruct, FLAN v2
Methodology:
Quantized Low-Rank Adapter (QLoRA) tuning
Model Architecture:
LoRA adapters on top of LLaMA models
Input Output 
Input Format:
Expected as text prompts formatted for dialogue.
Accepted Modalities:
text
Output Format:
Textual responses as a language model.
Performance Tips:
Loading in 16-bit may improve inference speed. Active development for efficient 4-bit inference kernels.
LLM NameQlora Self Instruct 65B
Repository 🤗https://huggingface.co/timdettmers/qlora-self-instruct-65b 
Model Size65b
Required VRAM3.2 GB
Updated2026-06-26
Maintainertimdettmers
Instruction-BasedYes
Model Files  3.2 GB
Model ArchitectureAdapter
Is Biasednone
PEFT TypeLORA
LoRA ModelYes
PEFT Target Modulesgate_proj|down_proj|up_proj|q_proj|k_proj|v_proj|o_proj
LoRA Alpha16
LoRA Dropout0.05
R Param64