INDEX

Explanations

phrases related to technology and coding, possibly focusing on specific terms or commands

oai_token-act-pair · gpt-3.5-turbo

specific identifiers, possibly related to names, titles, or character roles within content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

\<

-0.77

natureconservancy

-0.72

lehem

-0.68

etheless

-0.65

 gloss

-0.63

nown

-0.58

 slender

-0.57

 reversible

-0.57

 presidents

-0.56

abase

-0.56

POSITIVE LOGITS

backer

1.05

issance

0.91

kefeller

0.85

vous

0.84

restling

0.81

ctive

0.80

CTR

0.80

andom

0.76

ordan

0.75

 Grac

0.74

Activations Density 0.521%

phrases related to technology and coding, possibly focusing on specific terms or commands

specific identifiers, possibly related to names, titles, or character roles within content

No Comments

No Known Activations