INDEX

Explanations

names consisting of two words, with the second word being 'Wang'

oai_token-act-pair · gpt-3.5-turbo

mentions of the name "Wang."

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

phia

-0.76

TAIN

-0.74

ctic

-0.67

tarians

-0.66

cel

-0.65

 charge

-0.62

phis

-0.62

judicial

-0.62

 sucker

-0.61

 unch

-0.60

POSITIVE LOGITS

wana

1.06

enegger

1.02

 Chao

0.90

atu

0.86

aii

0.82

arro

0.80

orst

0.80

atche

0.78

ako

0.78

ata

0.77

Activations Density 0.018%

names consisting of two words, with the second word being 'Wang'

mentions of the name "Wang."

No Comments

No Known Activations