INDEX

Explanations

the word "Through" followed by a number

oai_token-act-pair · gpt-3.5-turbo

references to time periods or sequences

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

CTV

-0.72

juven

-0.70

abet

-0.68

ailable

-0.66

 whales

-0.64

SPA

-0.64

istic

-0.63

 inval

-0.63

etus

-0.62

IST

-0.62

POSITIVE LOGITS

lace

0.84

put

0.81

 Seasons

0.73

shaw

0.72

lik

0.69

through

0.68

drive

0.67

 intermedi

0.65

gression

0.65

 Rodgers

0.64

Activations Density 0.045%

the word "Through" followed by a number

references to time periods or sequences

No Comments

No Known Activations