INDEX

Explanations

the word "Trust"

oai_token-act-pair · gpt-3.5-turbo

references to trust-related entities and organizations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ĺħ

-0.75

 Centauri

-0.67

pmwiki

-0.65

GN

-0.62

cale

-0.62

issance

-0.61

OPLE

-0.61

Cue

-0.61

 Gale

-0.60

 Tone

-0.60

POSITIVE LOGITS

ees

1.52

worthiness

1.38

ee

1.19

 fund

1.02

worthy

1.00

fund

0.92

ieth

0.87

ishes

0.86

ulent

0.85

ership

0.85

Activations Density 0.026%

the word "Trust"

references to trust-related entities and organizations

No Comments

No Known Activations