INDEX

Explanations

instances of the word "pay"

oai_token-act-pair · gpt-3.5-turbo

references to the concept of paying attention or paying a price

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Scotia

-0.77

_>

-0.72

ember

-0.72

swick

-0.71

ãĤ¬

-0.70

ierre

-0.66

\/\/

-0.65

 Bend

-0.65

 Plain

-0.64

enburg

-0.64

POSITIVE LOGITS

checks

0.98

 heed

0.93

paying

0.91

loads

0.91

 dues

0.90

 attention

0.86

 homage

0.83

load

0.82

 paid

0.81

paid

0.80

Activations Density 0.035%

instances of the word "pay"

references to the concept of paying attention or paying a price

No Comments

No Known Activations