INDEX

Explanations

references to things or actions being incorrect, inappropriate, or harmful

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Configuration

neuronpedia/gpt2-small__res_scl-ajt/6-res_scl-ajt

Prompts (Dashboard)

12,288 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

46,080

Data Type

torch.float32

Hook Point

blocks.6.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

apollo-research/Skylion007-openwebtext-tokenizer-gpt2

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

casters

-0.73

anned

-0.72

ranging

-0.68

ivism

-0.68

lishes

-0.67

ury

-0.67

thood

-0.66

rs

-0.66

cit

-0.64

zeb

-0.63

POSITIVE LOGITS

 amount

0.90

 side

0.87

way

0.86

 kind

0.84

 thing

0.83

 piece

0.75

 direction

0.75

 solution

0.73

 person

0.72

 number

0.72

Activations Density 6.764%

references to things or actions being incorrect, inappropriate, or harmful

No Comments

No Known Activations