INDEX

Explanations

words related to hostile actions or behaviors

oai_token-act-pair · gpt-3.5-turbo

instances of the word "aggression" and its related forms

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

aver

-0.84

abol

-0.74

lev

-0.73

NCT

-0.72

ectar

-0.70

âĢ¢âĢ¢

-0.70

FORMATION

-0.69

view

-0.67

rote

-0.67

orah

-0.66

POSITIVE LOGITS

 aggression

1.36

 aggress

1.10

 provocation

1.01

Agg

0.93

 escalation

0.86

 aggressively

0.82

iveness

0.80

 towards

0.80

 retaliation

0.77

 escalate

0.77

Activations Density 0.015%

words related to hostile actions or behaviors

instances of the word "aggression" and its related forms

No Comments

No Known Activations