INDEX

Explanations

the word "steel" at different levels of relevance, with some activations indicating a very strong match

oai_token-act-pair · gpt-3.5-turbo

references to steel

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Kard

-0.76

 Niet

-0.74

 Garr

-0.73

romeda

-0.72

DOE

-0.69

[|

-0.67

itia

-0.66

 Chomsky

-0.64

uate

-0.64

annah

-0.63

POSITIVE LOGITS

works

1.06

 wool

1.05

Series

1.03

workers

1.00

worker

0.96

steel

0.94

anguage

0.91

fish

0.91

 beams

0.87

Steel

0.87

Activations Density 0.020%

the word "steel" at different levels of relevance, with some activations indicating a very strong match

references to steel

No Comments

No Known Activations