INDEX

Explanations

short phrases or sentences expressing physical actions or states

oai_token-act-pair · gpt-3.5-turbo

negative comments or criticisms about individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

iosyncr

-0.67

uates

-0.66

osponsors

-0.66

SpaceEngineers

-0.65

 ongoing

-0.61

 conduc

-0.60

erenn

-0.58

 coincides

-0.57

 Flavoring

-0.56

 preliminary

-0.56

POSITIVE LOGITS

he

1.20

He

1.20

Was

1.10

 Didn

1.08

He

1.07

didn

1.06

he

1.02

 didnt

1.02

Had

1.02

his

0.98

Activations Density 0.537%

short phrases or sentences expressing physical actions or states

negative comments or criticisms about individuals

No Comments

No Known Activations