INDEX

Explanations

discussions about confidential or off-the-record information

oai_token-act-pair · gpt-3.5-turbo

phrases related to political dynamics and decision-making processes

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

 murderer

-0.80

 Genocide

-0.79

 rapist

-0.74

ocide

-0.74

 perpetrator

-0.74

 fraudulent

-0.72

 horrendous

-0.71

 negro

-0.71

 USSR

-0.70

 horrend

-0.70

POSITIVE LOGITS

 aides

1.45

 POLITICO

1.30

 advisers

1.20

 confid

1.06

 briefed

1.05

ebus

1.04

 signaled

1.02

 Priebus

1.00

 strateg

1.00

 Kushner

0.97

Activations Density 0.887%

discussions about confidential or off-the-record information

phrases related to political dynamics and decision-making processes

No Comments

No Known Activations