INDEX

Explanations

the word "ress" followed by a high activation value, particularly "Tress"

oai_token-act-pair · gpt-3.5-turbo

instances of the word "press" and its variations

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

©¶æ

-0.83

ãĥ£

-0.80

£ı

-0.75

vernment

-0.68

rily

-0.66

 subp

-0.65

prus

-0.64

 volunte

-0.64

lder

-0.63

 hemisphere

-0.62

POSITIVE LOGITS

ively

1.06

ions

1.05

ional

0.99

encer

0.91

ants

0.91

IVE

0.89

entials

0.86

entially

0.86

ives

0.86

mann

0.86

Activations Density 0.018%

the word "ress" followed by a high activation value, particularly "Tress"

instances of the word "press" and its variations

No Comments

No Known Activations