INDEX

Explanations

numerical values related to combination or total

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate combined quantities or totals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

writers

-0.89

holes

-0.80

yers

-0.77

 Literary

-0.75

keeper

-0.75

guards

-0.73

otle

-0.72

der

-0.71

posts

-0.71

anty

-0.70

POSITIVE LOGITS

ãĤ¤ãĥĪ

0.76

ively

0.72

essions

0.71

ivia

0.70

ilde

0.70

 totals

0.68

 digits

0.67

sum

0.67

ogether

0.67

 sexes

0.66

Activations Density 0.017%

numerical values related to combination or total

phrases that indicate combined quantities or totals

No Comments

No Known Activations