INDEX

Explanations

names of persons, particularly "Denis" and "Dalton"

oai_token-act-pair · gpt-3.5-turbo

mentions of specific individuals, particularly those named Denis

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

taboola

-0.93

lished

-0.93

ramid

-0.90

 fixme

-0.84

natureconservancy

-0.79

rocal

-0.78

Assembly

-0.77

dri

-0.77

inventoryQuantity

-0.76

edded

-0.76

POSITIVE LOGITS

ovan

1.15

 Denis

0.93

ova

0.81

omorph

0.80

 Grimes

0.69

 Kats

0.69

 Congo

0.69

Mou

0.68

ovo

0.67

Wol

0.66

Activations Density 0.012%

names of persons, particularly "Denis" and "Dalton"

mentions of specific individuals, particularly those named Denis

No Comments

No Known Activations