INDEX

Explanations

words that describe actions that are done differently compared to the usual way

oai_token-act-pair · gpt-3.5-turbo

prepositions indicating alternatives or contrasts

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

hend

-0.75

isf

-0.71

iband

-0.71

izens

-0.70

CSS

-0.68

Ã£

-0.67

OSP

-0.67

Age

-0.67

Aren

-0.66

enary

-0.64

POSITIVE LOGITS

 necessarily

0.98

 ours

0.91

 anything

0.86

 theirs

0.84

 relying

0.82

 bothering

0.81

 outright

0.78

 usual

0.74

 ordinary

0.71

 hers

0.71

Activations Density 0.078%

words that describe actions that are done differently compared to the usual way

prepositions indicating alternatives or contrasts

No Comments

No Known Activations