INDEX

Explanations

phrases related to numerical expressions and statistics

oai_token-act-pair · gpt-3.5-turbo

references to numeric values and their contexts

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

 Till

-0.65

 Rece

-0.64

 Zionism

-0.64

 Fritz

-0.64

 Shank

-0.63

 Azerb

-0.63

 Kodi

-0.63

 Turkey

-0.62

 Vega

-0.62

 Goods

-0.62

POSITIVE LOGITS

apixel

0.80

 fists

0.76

imensional

0.75

66666666

0.74

 veto

0.70

GHz

0.69

oz

0.68

 handc

0.68

margin

0.66

 fist

0.66

Activations Density 0.249%

phrases related to numerical expressions and statistics

references to numeric values and their contexts

No Comments

No Known Activations