INDEX

Explanations

references to past incidents or events

oai_token-act-pair · gpt-3.5-turbo

phrases related to investigations and disclosures of information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 traveler

-0.80

colored

-0.80

avored

-0.79

Favorite

-0.76

 civilizations

-0.76

 civilization

-0.75

 capitalize

-0.74

 mustache

-0.74

Hispanic

-0.73

 traveling

-0.73

POSITIVE LOGITS

 Â£

1.18

 criticised

1.17

MPs

1.16

 whistleblower

1.15

 councillors

1.09

 breaches

1.09

 whistleblowers

1.09

NHS

1.04

 revelations

1.03

 failings

1.02

Activations Density 0.555%

references to past incidents or events

phrases related to investigations and disclosures of information

No Comments

No Known Activations