INDEX

Explanations

phrases indicating evaluation or contemplation of information

oai_token-act-pair · gpt-3.5-turbo

instances of the word "considering."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ode

-0.60

atre

-0.57

orc

-0.56

eva

-0.55

 shuttle

-0.55

 crashed

-0.54

cade

-0.54

 rows

-0.53

ater

-0.53

 binds

-0.53

POSITIVE LOGITS

 considering

3.46

 contemplating

1.91

 Considering

1.68

Considering

1.65

consider

1.65

 evaluating

1.41

 reconsider

1.37

 comparing

1.37

 judging

1.37

 consideration

1.35

Activations Density 0.015%

phrases indicating evaluation or contemplation of information

instances of the word "considering."

No Comments

No Known Activations