INDEX

Explanations

the word "winner" with various intensities

oai_token-act-pair · gpt-3.5-turbo

terms related to winners and winning outcomes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

uli

-0.75

uku

-0.64

Dub

-0.63

aeda

-0.63

abb

-0.62

uilt

-0.62

rients

-0.62

Ku

-0.61

ORE

-0.60

ugu

-0.60

POSITIVE LOGITS

 winner

1.19

 Winner

1.05

Winner

1.05

 winners

0.97

 contestant

0.95

 prize

0.88

 laureate

0.88

 loser

0.88

winner

0.84

stakes

0.79

Activations Density 0.016%

the word "winner" with various intensities

terms related to winners and winning outcomes

No Comments

No Known Activations