INDEX

Explanations

references to people with the title "President."

oai_token-act-pair · gpt-3.5-turbo

references to a specific individual or character named "Pr."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.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

 EntityItem

-0.75

rities

-0.67

INESS

-0.66

dayName

-0.66

ãĥ¼ãĥ«

-0.65

RAFT

-0.64

beard

-0.63

soever

-0.63

 localization

-0.62

 spirited

-0.61

POSITIVE LOGITS

atche

1.27

udence

1.18

imes

1.04

ima

1.00

ussia

0.98

inter

0.98

icer

0.98

ices

0.98

acking

0.97

ussian

0.96

Activations Density 0.018%

references to people with the title "President."

references to a specific individual or character named "Pr."

No Comments

No Known Activations