INDEX

Explanations

the word "anything" with high activation values

oai_token-act-pair · gpt-3.5-turbo

the term "anything" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

irth

-0.75

 anonymity

-0.68

 Examiner

-0.67

iger

-0.67

oliath

-0.66

ilt

-0.66

orf

-0.64

pu

-0.64

ruck

-0.64

fer

-0.64

POSITIVE LOGITS

 else

1.71

 resembling

1.36

Else

1.32

 Else

1.08

 remotely

1.07

else

0.94

THING

0.92

 whatsoever

0.91

 happening

0.90

soever

0.85

Activations Density 0.059%

the word "anything" with high activation values

the term "anything" in various contexts

No Comments

No Known Activations