INDEX

Explanations

phrases related to articles or submissions, possibly with a call to action or donation request

oai_token-act-pair · gpt-3.5-turbo

instances of strong emotional expressions and reactions

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

 courier

-0.70

 Aval

-0.69

 herself

-0.68

 intended

-0.67

escription

-0.67

 intend

-0.66

 sacked

-0.65

 uninterrupted

-0.65

 estranged

-0.64

 compr

-0.64

POSITIVE LOGITS

Anyway

1.46

 Anyway

1.10

Seriously

1.07

Advertisement

1.07

Consider

1.01

Unless

0.98

Here

0.95

Of

0.94

fuck

0.94

Speaking

0.93

Activations Density 0.661%

phrases related to articles or submissions, possibly with a call to action or donation request

instances of strong emotional expressions and reactions

No Comments

No Known Activations