INDEX

Explanations

nouns representing companies or organizations

oai_token-act-pair · gpt-3.5-turbo

phrases indicating surprising or significant comparisons

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 sketches

-0.75

reci

-0.74

iece

-0.69

 ailments

-0.67

 cliff

-0.67

enei

-0.67

livion

-0.66

 reversible

-0.66

 wrinkles

-0.66

ursday

-0.66

POSITIVE LOGITS

 ISPs

1.12

Companies

1.11

 organisations

1.07

 Comcast

1.07

 Cosponsors

1.06

 Alibaba

1.06

 Coca

1.05

 Deutsche

1.03

 Companies

1.02

 HSBC

1.02

Activations Density 0.527%

nouns representing companies or organizations

phrases indicating surprising or significant comparisons

No Comments

No Known Activations