INDEX

Explanations

phrases or words related to forceful and aggressive actions, especially verbal attacks or criticisms

oai_token-act-pair · gpt-3.5-turbo

terms related to corporal punishment or physical reprimand

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Eugene

-0.86

 Mystery

-0.80

UC

-0.78

 Meier

-0.74

UC

-0.72

uria

-0.71

 Cipher

-0.71

 Gravity

-0.71

 Greens

-0.71

UCS

-0.71

POSITIVE LOGITS

 lash

3.60

 lashes

3.34

 lashed

2.66

 eyel

1.88

 Lash

1.77

 lamb

1.61

lash

1.48

 mascara

1.14

lain

1.03

lob

1.01

Activations Density 0.036%

phrases or words related to forceful and aggressive actions, especially verbal attacks or criticisms

terms related to corporal punishment or physical reprimand

No Comments

No Known Activations