INDEX

Explanations

blobs of unusual text patterns that do not seem to be related to any specific theme

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

schild

-0.88

ATTLE

-0.87

clerosis

-0.85

ALLY

-0.80

 Whole

-0.79

 Clover

-0.77

 Sabha

-0.76

 rapt

-0.76

 Anxiety

-0.75

 voic

-0.74

POSITIVE LOGITS

ongyang

1.46

asus

1.30

lator

1.19

Cola

1.17

merga

1.13

formance

1.12

eworthy

1.12

sylvania

1.11

cipled

1.11

chedel

1.05

Activations Density 15.369%

blobs of unusual text patterns that do not seem to be related to any specific theme

No Comments

No Known Activations