INDEX

Explanations

numbers in the format of thousands (e.g., 000) with high activation levels

oai_token-act-pair · gpt-3.5-turbo

numeric values expressed in thousands

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

swick

-0.78

warts

-0.78

illard

-0.73

krit

-0.73

riot

-0.71

icka

-0.69

vironment

-0.69

agall

-0.67

 weakness

-0.66

vation

-0.66

POSITIVE LOGITS

0.81

Mbps

0.81

 è£ıè¦ļéĨĴ

0.75

Hz

0.74

ãĥ©ãĥ³

0.72

Hz

0.70

mAh

0.70

HT

0.70

0.69

Activations Density 0.061%

numbers in the format of thousands (e.g., 000) with high activation levels

numeric values expressed in thousands

No Comments

No Known Activations