INDEX

Explanations

actions or activities related to checking or monitoring

oai_token-act-pair · gpt-3.5-turbo

instances related to daily life activities and interactions

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

 underestimated

-0.56

 inherently

-0.55

 Seller

-0.51

 unfairly

-0.51

Pwr

-0.49

 nowadays

-0.49

 inherent

-0.49

 depends

-0.48

 historically

-0.48

 reliant

-0.48

POSITIVE LOGITS

romptu

0.64

clock

0.56

USS

0.56

morning

0.54

cellaneous

0.53

aturdays

0.53

FIR

0.52

 mosqu

0.52

 chanting

0.52

 benches

0.52

Activations Density 1.721%

actions or activities related to checking or monitoring

instances related to daily life activities and interactions

No Comments

No Known Activations