INDEX

Explanations

instances of the word "notices"

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "notices."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 characterized

-0.67

idis

-0.66

lege

-0.65

 compromise

-0.62

 foundation

-0.61

actionDate

-0.61

 normalized

-0.61

dyn

-0.61

 distort

-0.61

 simmer

-0.60

POSITIVE LOGITS

 notices

3.17

 announcements

1.45

 slips

1.44

 notifications

1.43

 replies

1.42

 mentions

1.42

 summons

1.31

 asks

1.31

 alerts

1.30

 tags

1.26

Activations Density 0.036%

instances of the word "notices"

occurrences of the word "notices."

No Comments

No Known Activations