INDEX

Explanations

proper nouns related to the name "Burn" or similar patterns

oai_token-act-pair · gpt-3.5-turbo

mentions of the word "burn."

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

oteric

-0.69

atro

-0.64

Magikarp

-0.64

 abusive

-0.64

 unethical

-0.62

 unfor

-0.60

 oblig

-0.60

INAL

-0.59

 dime

-0.59

 Americ

-0.58

POSITIVE LOGITS

burn

1.08

er

0.97

din

0.90

hardt

0.89

ished

0.89

robe

0.87

ards

0.87

sburg

0.83

erate

0.82

ed

0.81

Activations Density 0.013%

proper nouns related to the name "Burn" or similar patterns

mentions of the word "burn."

No Comments

No Known Activations