INDEX

Explanations

phrases related to expressing strong opinions or critiques

oai_token-act-pair · gpt-3.5-turbo

phrases related to controversial political issues

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

 Processing

-0.86

 Twins

-0.76

 CrossRef

-0.70

ITNESS

-0.67

 Procedures

-0.66

luck

-0.65

 Caption

-0.65

 Composite

-0.64

 Browse

-0.64

POSITIVE LOGITS

 vehemently

1.21

 bashing

1.17

 denouncing

1.13

 anti

1.11

 condemn

1.11

 denounce

1.10

 opposing

1.10

 criticizing

1.08

 staunch

1.08

 advoc

1.07

Activations Density 0.502%

phrases related to expressing strong opinions or critiques

phrases related to controversial political issues

No Comments

No Known Activations