INDEX

Explanations

It seems like the neuron is mainly looking for words related to negation

oai_token-act-pair · gpt-3.5-turbo

the word "not" and its variations in context

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

stakes

-0.71

 Circuit

-0.71

itor

-0.70

 Spotlight

-0.69

 Tycoon

-0.67

 Pros

-0.67

 Contrast

-0.65

 Comparison

-0.65

 Expansion

-0.65

 Handbook

-0.64

POSITIVE LOGITS

icably

1.39

epad

1.20

icable

1.15

 necessarily

1.10

hin

1.04

orious

0.95

ched

0.92

withstanding

0.89

yet

0.85

ifications

0.83

Activations Density 0.171%

It seems like the neuron is mainly looking for words related to negation

the word "not" and its variations in context

No Comments

No Known Activations