INDEX

Explanations

instances where someone or something is implicated or involved in a negative event or action

oai_token-act-pair · gpt-3.5-turbo

verbs indicating former actions or states related to allegations and involvement in events

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

ipeg

-0.85

iann

-0.83

 Adds

-0.71

 Starts

-0.71

ggles

-0.70

ategor

-0.70

izable

-0.69

 adjective

-0.68

inav

-0.67

ilight

-0.67

POSITIVE LOGITS

 involved

1.19

 responsible

1.15

 complicit

1.13

 intoxicated

1.10

 guilty

1.07

 negligent

1.07

 culp

1.06

 aware

1.04

 plotting

1.03

 abusing

1.01

Activations Density 0.168%

instances where someone or something is implicated or involved in a negative event or action

verbs indicating former actions or states related to allegations and involvement in events

No Comments

No Known Activations