INDEX

Explanations

mentions of future actions by individuals

oai_token-act-pair · gpt-3.5-turbo

instances of the word "would" indicating habitual actions or future intentions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Corpus

-0.65

rylic

-0.65

resources

-0.65

 Efficiency

-0.65

 Gutenberg

-0.64

values

-0.64

advant

-0.62

effic

-0.62

Kag

-0.59

ractical

-0.59

POSITIVE LOGITS

be

1.00

 doubtless

0.96

 undoubtedly

0.89

 gladly

0.89

 dearly

0.88

 never

0.85

 eventually

0.85

ĸļ

0.83

 surely

0.82

've

0.79

Activations Density 0.145%

mentions of future actions by individuals

instances of the word "would" indicating habitual actions or future intentions

No Comments

No Known Activations