INDEX

Explanations

references to strong negative emotions or concerns

oai_token-act-pair · gpt-3.5-turbo

instances of the word "fears" and its associated context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ced

-0.77

thumbnails

-0.75

arb

-0.74

nice

-0.74

urgy

-0.74

sample

-0.70

lease

-0.69

ann

-0.69

ergy

-0.68

cise

-0.68

POSITIVE LOGITS

 fears

1.20

wart

0.87

lessly

0.87

 lest

0.81

 worries

0.75

warts

0.75

 perceptions

0.74

 fear

0.73

mong

0.73

 fearing

0.72

Activations Density 0.007%

references to strong negative emotions or concerns

instances of the word "fears" and its associated context

No Comments

No Known Activations