INDEX

Explanations

contractions where 'am' or 'I'm' is used

oai_token-act-pair · gpt-3.5-turbo

expressions of personal opinion or self-referential statements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

glers

-0.74

 externalToEVAOnly

-0.68

eness

-0.68

 Dynamics

-0.65

Gap

-0.64

å£«

-0.63

aneers

-0.63

Prev

-0.62

illance

-0.62

aceutical

-0.61

POSITIVE LOGITS

 glad

1.32

 guessing

1.30

 sure

1.24

 amazed

1.21

 sorry

1.20

 afraid

1.14

 thankful

1.11

 tempted

1.11

 ashamed

1.09

 grateful

1.08

Activations Density 0.134%

contractions where 'am' or 'I'm' is used

expressions of personal opinion or self-referential statements

No Comments

No Known Activations