INDEX

Explanations

accounts or statements made by spokespersons

oai_token-act-pair · gpt-3.5-turbo

the word "spokesman" and its variations in reporting contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

edu

-0.77

Reward

-0.73

 gorge

-0.72

notations

-0.71

llah

-0.70

 repaid

-0.69

ptions

-0.67

venants

-0.65

ibe

-0.64

hu

-0.64

POSITIVE LOGITS

bidden

0.96

 instance

0.82

 example

0.81

gotten

0.80

 STATS

0.78

cing

0.78

 managing

0.78

cers

0.77

 Sierra

0.76

 defending

0.75

Activations Density 0.105%

accounts or statements made by spokespersons

the word "spokesman" and its variations in reporting contexts

No Comments

No Known Activations