|
gptkbp:instance_of
|
gptkb:Artificial_Intelligence
|
|
gptkbp:allows
|
Context length
Factual inaccuracy
Sensitivity to input phrasing
|
|
gptkbp:application
|
gptkb:software
|
|
gptkbp:business_model
|
Bias in outputs
Misinformation generation
|
|
gptkbp:community_support
|
Active community
|
|
gptkbp:competitors
|
gptkb:BERT
gptkb:T5
XL Net
|
|
gptkbp:developed_by
|
gptkb:Open_AI
|
|
gptkbp:first_released
|
June 2018
|
|
gptkbp:has_ability
|
Question answering
Text generation
Language translation
Text completion
|
|
https://www.w3.org/2000/01/rdf-schema#label
|
GPT Series
|
|
gptkbp:impact
|
gptkb:Significant
|
|
gptkbp:input_output
|
Text prompts
Text responses
|
|
gptkbp:is_evaluated_by
|
Perplexity
BLEU score
|
|
gptkbp:is_integrated_with
|
gptkb:battle
|
|
gptkbp:is_open_source
|
Partially
|
|
gptkbp:language
|
Multiple languages
Generative Pre-trained Transformer
|
|
gptkbp:latest_version
|
gptkb:GPT-4
|
|
gptkbp:marketed_as
|
gptkb:battle
|
|
gptkbp:notable_features
|
Few-shot learning
Interactive dialogue
Zero-shot learning
Fine-tuning capabilities
|
|
gptkbp:performance
|
State-of-the-art
|
|
gptkbp:project
|
gptkb:legislation
Ongoing research
Model improvements
|
|
gptkbp:provides_information_on
|
Diverse internet text
|
|
gptkbp:related_model
|
Varies by version
|
|
gptkbp:system_requirements
|
High-performance GP Us
|
|
gptkbp:training
|
Weeks to months
Unsupervised learning
|
|
gptkbp:tuning
|
Possible
|
|
gptkbp:user_base
|
Developers and researchers
|
|
gptkbp:user_interface
|
API and web interface
|
|
gptkbp:uses
|
Transformer architecture
|
|
gptkbp:bfsParent
|
gptkb:GPT-1
|
|
gptkbp:bfsLayer
|
5
|