INDEX

Explanations

news headlines indicating a location

oai_token-act-pair · gpt-3.5-turbo

parentheses in the text

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

 expire

-0.71

 overlap

-0.71

racuse

-0.68

 surround

-0.67

 glare

-0.66

 retard

-0.66

 majesty

-0.66

 entitle

-0.64

 multiplication

-0.64

 sonic

-0.63

POSITIVE LOGITS

Reuters

1.17

MEN

1.14

via

1.14

formerly

1.09

CBS

1.08

pictured

1.03

AFP

1.03

TAG

1.02

BUS

0.99

...)

0.97

Activations Density 0.050%

news headlines indicating a location

parentheses in the text

No Comments

No Known Activations