INDEX

Explanations

mentions of things that are very hot

oai_token-act-pair · gpt-3.5-turbo

instances of the word "hot" used in various contexts

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

eca

-0.76

ACA

-0.75

arnaev

-0.75

ARDIS

-0.73

ufact

-0.73

aughs

-0.73

ISION

-0.73

vous

-0.71

IX

-0.69

 Borders

-0.69

POSITIVE LOGITS

hot

1.13

hot

1.07

 hotter

1.04

 hottest

1.04

HOT

0.94

 hots

0.91

 sauce

0.85

Hot

0.85

Hot

0.84

 spots

0.84

Activations Density 0.011%

mentions of things that are very hot

instances of the word "hot" used in various contexts

No Comments

No Known Activations