INDEX

Explanations

phrases related to observation or visual perception

oai_token-act-pair · gpt-3.5-turbo

instances of the verb "see."

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

lessly

-0.65

lain

-0.64

rained

-0.60

onson

-0.59

itudes

-0.58

ongh

-0.57

han

-0.57

wrapper

-0.55

lete

-0.55

acted

-0.53

POSITIVE LOGITS

see

3.17

See

2.00

see

1.89

SEE

1.84

See

1.82

 sees

1.80

 hear

1.69

saw

1.56

 seeing

1.53

 perceive

1.50

Activations Density 0.057%

phrases related to observation or visual perception

instances of the verb "see."

No Comments

No Known Activations