INDEX

Explanations

words related to errors or problems

oai_token-act-pair · gpt-3.5-turbo

references to flaws and weaknesses in systems or concepts

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

 Carbuncle

-0.72

 Jackets

-0.69

NAS

-0.67

zee

-0.66

rontal

-0.66

rollers

-0.65

xon

-0.63

NetMessage

-0.63

asty

-0.63

Pain

-0.62

POSITIVE LOGITS

lessly

1.06

acies

0.91

 flaws

0.87

 flaw

0.83

 plag

0.81

ulence

0.81

nah

0.73

lessness

0.72

 compromises

0.72

ously

0.72

Activations Density 0.031%

words related to errors or problems

references to flaws and weaknesses in systems or concepts

No Comments

No Known Activations