INDEX

Explanations

proper nouns and titles

oai_token-act-pair · gpt-3.5-turbo

instances of direct statements or assertions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

Enlarge

-1.26

-'

-0.79

Flickr

-0.76

'-

-0.76

Via

-0.74

Exit

-0.73

ono

-0.72

Ct

-0.71

toggle

-0.70

POSITIVE LOGITS

 cannot

1.13

 will

0.83

 must

0.79

 learnt

0.76

 cant

0.76

can

0.75

may

0.74

is

0.73

 conclud

0.72

 envis

0.71

Activations Density 0.410%

proper nouns and titles

instances of direct statements or assertions

No Comments

No Known Activations