INDEX

Explanations

words associated with specific concepts or attributes

oai_token-act-pair · gpt-3.5-turbo

phrases that reference associations with various concepts or phenomena

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

chal

-0.80

stall

-0.77

ardless

-0.75

hene

-0.74

annis

-0.73

ppe

-0.72

umblr

-0.71

ceans

-0.68

dan

-0.67

vette

-0.66

POSITIVE LOGITS

ively

0.88

 affili

0.83

 newsp

0.81

 associations

0.75

unct

0.75

ually

0.75

 activ

0.74

lia

0.74

vertisements

0.73

umni

0.71

Activations Density 0.014%

words associated with specific concepts or attributes

phrases that reference associations with various concepts or phenomena

No Comments

No Known Activations