INDEX

Explanations

mentions of legislation or political actions

oai_token-act-pair · gpt-3.5-turbo

mentions of lenses or visual perspectives

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 mathemat

-0.86

 incorpor

-0.78

 lawy

-0.70

 contrace

-0.68

 predec

-0.67

 pyramid

-0.67

 fortun

-0.66

 unden

-0.64

 opportunities

-0.64

 welf

-0.64

POSITIVE LOGITS

ï¸ı

1.77

ï¸

1.05

Â£

0.96

âĻ

0.89

ski

0.86

âĶĢâĶĢâĶĢâĶĢ

0.85

âĶĢâĶĢ

0.84

Â·Â·

0.83

Â¯Â¯

0.82

stellar

0.81

Activations Density 0.264%

mentions of legislation or political actions

mentions of lenses or visual perspectives

No Comments

No Known Activations