INDEX

Explanations

high-ranking positions or status

oai_token-act-pair · gpt-3.5-turbo

references to "high" in various contexts, particularly in terms of status or intensity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ĸļ

-0.82

tnc

-0.73

velength

-0.66

vous

-0.66

unia

-0.65

undo

-0.62

uphem

-0.61

okemon

-0.59

 Unlimited

-0.58

near

-0.58

POSITIVE LOGITS

profile

1.01

brow

0.99

 stakes

0.98

fal

0.97

 profile

0.96

ranking

0.95

 ranking

0.92

landers

0.92

rollers

0.86

stakes

0.85

Activations Density 0.077%

high-ranking positions or status

references to "high" in various contexts, particularly in terms of status or intensity

No Comments

No Known Activations