INDEX

Explanations

locations of news events

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 tempt

-0.89

 renegoti

-0.88

 boycot

-0.85

 monopol

-0.85

 challeng

-0.85

 concess

-0.81

 integ

-0.81

 transact

-0.77

 embargo

-0.77

 microbiome

-0.77

POSITIVE LOGITS

 Witnesses

1.38

 Investigators

1.28

 Authorities

1.25

 Officials

1.14

 Dozens

1.14

 Luckily

1.13

 Thankfully

1.11

Authorities

1.10

 Fortunately

1.10

Witness

1.09

Activations Density 0.202%

locations of news events

No Comments

No Known Activations