INDEX

Explanations

instances of the word "whom."

oai_token-act-pair · gpt-3.5-turbo

references to individuals or groups

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Loading

-0.74

âĨ

-0.71

reach

-0.68

pad

-0.66

 Delicious

-0.65

 Belt

-0.64

aster

-0.62

Charg

-0.62

cluding

-0.62

belt

-0.62

POSITIVE LOGITS

soever

1.99

odox

0.79

awei

0.78

izens

0.78

umbnails

0.78

owship

0.74

vou

0.73

igham

0.72

omever

0.71

itsu

0.71

Activations Density 0.012%

instances of the word "whom."

references to individuals or groups

No Comments

No Known Activations