INDEX

Explanations

The neuron is searching for specific text patterns or phrases beginning with symbols like 'âĢ'

oai_token-act-pair · gpt-3.5-turbo

instances of repeated phrases or placeholders in a text

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

 Kenyan

-0.74

vt

-0.69

 Mansion

-0.67

liest

-0.64

lim

-0.63

 Saga

-0.62

 Recovery

-0.62

 Trip

-0.60

è£ħ

-0.60

Mun

-0.60

POSITIVE LOGITS

£

0.88

º

0.87

âģ

0.81

¹

0.80

Ì

0.79

abad

0.79

âĹ¼

0.77

acca

0.74

¬

0.72

Pg

0.70

Activations Density 0.036%

The neuron is searching for specific text patterns or phrases beginning with symbols like 'âĢ'

instances of repeated phrases or placeholders in a text

No Comments

No Known Activations