INDEX

Explanations

words related to rankings or positions

oai_token-act-pair · gpt-3.5-turbo

references to ranked lists or top selections

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 Lauder

-0.64

 Gaul

-0.62

Cla

-0.61

 Baldwin

-0.60

otive

-0.60

ary

-0.60

ATIONS

-0.58

 Birth

-0.58

 Arri

-0.57

uctions

-0.57

POSITIVE LOGITS

ographical

1.16

most

1.09

ography

1.07

deck

1.04

ology

0.99

ographically

0.95

ographic

0.94

eka

0.92

 notch

0.91

ologically

0.90

Activations Density 0.045%

words related to rankings or positions

references to ranked lists or top selections

No Comments

No Known Activations