INDEX

Explanations

phrases related to understanding or justification

oai_token-act-pair · gpt-3.5-turbo

expressions of comprehensibility or rationality

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

orp

-0.65

fast

-0.63

illac

-0.62

EE

-0.62

rock

-0.60

oo

-0.59

 moon

-0.59

 Brass

-0.57

Â°

-0.57

TION

-0.56

POSITIVE LOGITS

 understandable

3.71

 manageable

1.50

 understandably

1.50

 admirable

1.43

 incomprehensible

1.36

 believable

1.29

 predictable

1.22

 palpable

1.15

 inexplicable

1.12

 readable

1.11

Activations Density 0.024%

phrases related to understanding or justification

expressions of comprehensibility or rationality

No Comments

No Known Activations