INDEX

Explanations

terms related to something remaining or staying the same

oai_token-act-pair · gpt-3.5-turbo

phrases indicating consistency or lack of change

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 McKenna

-0.66

ueller

-0.65

aph

-0.63

ologne

-0.60

ushes

-0.59

gee

-0.59

umper

-0.59

quickShipAvailable

-0.58

UNE

-0.58

 Lama

-0.57

POSITIVE LOGITS

 unchanged

0.94

 intact

0.80

iated

0.79

edly

0.77

lihood

0.77

enance

0.74

aneously

0.73

ledged

0.72

erella

0.72

ishment

0.70

Activations Density 0.030%

terms related to something remaining or staying the same

phrases indicating consistency or lack of change

No Comments

No Known Activations