INDEX

Explanations

ironic or contradictory statements in the text

oai_token-act-pair · gpt-3.5-turbo

concepts related to irony and moral complexity

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

 Picks

-0.67

 originals

-0.66

 Dreams

-0.63

 Hate

-0.57

 prefers

-0.56

 IMAGES

-0.55

 halves

-0.55

 Cheap

-0.55

 Occupations

-0.55

 lashes

-0.55

POSITIVE LOGITS

 involved

1.15

 brewing

1.07

 lurking

1.05

 underway

1.05

 happening

1.03

 attached

0.96

 here

0.95

 occurring

0.95

 going

0.92

 everywhere

0.92

Activations Density 0.253%

ironic or contradictory statements in the text

concepts related to irony and moral complexity

No Comments

No Known Activations