INDEX

Explanations

specific words related to watches

oai_token-act-pair · gpt-3.5-turbo

references to watches and watch-related terminology

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

Reviewer

-0.82

«ĺ

-0.78

EngineDebug

-0.76

bryce

-0.67

 congr

-0.65

lil

-0.65

¬¼

-0.65

++++++++++++++++

-0.64

å§«

-0.63

ãĤ¨ãĥ«

-0.63

POSITIVE LOGITS

tower

1.31

dogs

1.20

dog

1.08

watch

1.01

maker

0.91

 watches

0.88

bands

0.87

 Watching

0.86

strap

0.85

list

0.84

Activations Density 0.030%

specific words related to watches

references to watches and watch-related terminology

No Comments

No Known Activations