INDEX

Explanations

words related to breaking or failure

oai_token-act-pair · gpt-3.5-turbo

references to the term "bust" and its variations in different contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 vomit

-0.67

WAYS

-0.65

 Rouge

-0.64

 Cruel

-0.62

ised

-0.62

 Pradesh

-0.59

mble

-0.59

dfx

-0.58

ndra

-0.58

 vomiting

-0.57

POSITIVE LOGITS

le

0.95

lar

0.92

buster

0.90

aign

0.89

enegger

0.89

neck

0.89

0.88

cies

0.88

ards

0.87

les

0.87

Activations Density 0.039%

words related to breaking or failure

references to the term "bust" and its variations in different contexts

No Comments

No Known Activations