INDEX

Explanations

references to legal or governmental entities like courts and appointments

oai_token-act-pair · gpt-3.5-turbo

references to a specific topic or event related to scores and rankings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

éĹĺ

-0.74

 loose

-0.72

 prevail

-0.67

 household

-0.66

 restraint

-0.65

 dare

-0.65

 remembrance

-0.65

 liberating

-0.63

 farewell

-0.62

ãģį

-0.62

POSITIVE LOGITS

ratch

1.13

opes

1.13

attered

1.05

apers

1.03

atters

1.02

rupulous

1.02

urry

1.00

iple

1.00

reenshots

0.98

itizens

0.98

Activations Density 0.006%

references to legal or governmental entities like courts and appointments

references to a specific topic or event related to scores and rankings

No Comments

No Known Activations