INDEX

Explanations

numerical values indicating quantities

oai_token-act-pair · gpt-3.5-turbo

phrases indicating quantities or measurements slightly exceeding a specified amount

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

antine

-0.87

Ô

-0.74

cean

-0.74

-0.70

illary

-0.70

 Flavoring

-0.69

PLA

-0.69

ubi

-0.68

uld

-0.65

Creat

-0.64

POSITIVE LOGITS

 Pradesh

0.72

kill

0.68

population

0.67

 horizon

0.65

shy

0.63

whelming

0.63

 half

0.62

deck

0.62

akuya

0.62

capacity

0.61

Activations Density 0.030%

numerical values indicating quantities

phrases indicating quantities or measurements slightly exceeding a specified amount

No Comments

No Known Activations