INDEX

Explanations

words ending in 'ue'

oai_token-act-pair · gpt-3.5-turbo

occurrences of the substring "ue."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 decor

-0.66

 ultras

-0.65

 intr

-0.65

 defending

-0.64

 Benghazi

-0.63

 harvest

-0.63

 insure

-0.62

 upholding

-0.62

 crust

-0.61

 vibr

-0.59

POSITIVE LOGITS

ue

4.53

ues

2.85

ued

2.36

UE

2.29

uing

2.13

UES

1.74

uer

1.74

uel

1.68

uers

1.65

ua

1.33

Activations Density 0.011%

words ending in 'ue'

occurrences of the substring "ue."

No Comments

No Known Activations