INDEX

Explanations

references to geographical locations, particularly places like Washington

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "Washington"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

Rating

-0.69

ISTER

-0.69

etimes

-0.67

abol

-0.67

unct

-0.67

cffffcc

-0.66

Reloaded

-0.66

ONSORED

-0.65

displayText

-0.65

schild

-0.65

POSITIVE LOGITS

DC

1.04

 Capitals

1.02

 Redskins

1.01

 Monument

0.95

 Washington

0.94

 Dull

0.93

 Nationals

0.91

Washington

0.86

 Everett

0.83

 Bridge

0.82

Activations Density 0.020%

references to geographical locations, particularly places like Washington

occurrences of the word "Washington"

No Comments

No Known Activations