INDEX

Explanations

comparisons or quantifications related to power or influence

oai_token-act-pair · gpt-3.5-turbo

quantifiers that indicate different degrees or amounts of influence or power

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

xtap

-0.78

 Reporter

-0.77

aiman

-0.69

cade

-0.69

FIG

-0.68

 Digest

-0.68

jen

-0.68

duction

-0.67

abee

-0.67

hemer

-0.66

POSITIVE LOGITS

 clout

1.16

 patience

1.10

 sympathy

1.03

 doubts

1.03

 advantages

1.02

 trouble

1.02

 aspirations

1.01

 recourse

0.99

 responsibilities

0.99

 confidence

0.97

Activations Density 0.166%

comparisons or quantifications related to power or influence

quantifiers that indicate different degrees or amounts of influence or power

No Comments

No Known Activations