INDEX

Explanations

verbs related to causing harm

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "ruin" and its derivatives

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

appa

-0.74

duino

-0.68

heter

-0.67

enne

-0.67

bors

-0.67

leground

-0.67

arij

-0.66

gencies

-0.66

taboola

-0.63

WER

-0.63

POSITIVE LOGITS

 havoc

1.07

ous

0.96

ously

0.94

OUS

0.81

fully

0.81

stal

0.81

 spoil

0.79

spo

0.77

strument

0.76

ifully

0.76

Activations Density 0.018%

verbs related to causing harm

references to the concept of "ruin" and its derivatives

No Comments

No Known Activations