INDEX

Explanations

adjectives related to size or scale

oai_token-act-pair · gpt-3.5-turbo

terms related to size and scale, particularly "sprawling" and similar descriptors

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

[+

-0.77

*/(

-0.74

ighed

-0.65

activated

-0.65

yl

-0.64

saf

-0.64

utral

-0.63

AUT

-0.62

Evidence

-0.62

ordan

-0.62

POSITIVE LOGITS

 sprawling

1.23

 horizont

0.89

 expans

0.81

halla

0.80

 inland

0.80

 campus

0.76

hog

0.76

 adolesc

0.75

ishly

0.74

 swath

0.74

Activations Density 0.016%

adjectives related to size or scale

terms related to size and scale, particularly "sprawling" and similar descriptors

No Comments

No Known Activations