INDEX

Explanations

instances where someone expresses a desire not to do something or expresses a lack of want

oai_token-act-pair · gpt-3.5-turbo

instances of the phrase "to" followed by any verb

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Appears

-0.63

lip

-0.60

 assessments

-0.58

pointers

-0.58

 representations

-0.57

quickShipAvailable

-0.57

 depictions

-0.57

 surges

-0.57

 deployments

-0.56

tions

-0.55

POSITIVE LOGITS

be

1.07

 fulfill

1.05

 satisfy

1.04

 keep

1.03

 give

1.03

 accommodate

1.03

 minimize

1.03

 make

1.01

 maximize

0.99

 provide

0.99

Activations Density 1.113%

instances where someone expresses a desire not to do something or expresses a lack of want

instances of the phrase "to" followed by any verb

No Comments

No Known Activations