INDEX

Explanations

adjectives emphasizing a high degree of something

oai_token-act-pair · gpt-3.5-turbo

instances of the word "pretty" used to emphasize descriptions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

NER

-0.72

Footnote

-0.72

APD

-0.71

IAS

-0.69

ocene

-0.67

idia

-0.66

KER

-0.66

arta

-0.65

iment

-0.64

JV

-0.64

POSITIVE LOGITS

 darn

1.41

 much

1.21

 damn

1.09

 damned

1.05

 nifty

1.00

much

0.98

 MUCH

0.94

 straightforward

0.89

 lousy

0.86

 neat

0.86

Activations Density 0.027%

adjectives emphasizing a high degree of something

instances of the word "pretty" used to emphasize descriptions

No Comments

No Known Activations