INDEX

Explanations

occurrences of the term "Am" followed by a number

oai_token-act-pair · gpt-3.5-turbo

instances of the abbreviation "Am" followed by numbers, indicating references to specific entities or organizations

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

ãģį

-0.79

 corners

-0.73

theless

-0.71

 directions

-0.70

 edges

-0.67

 learners

-0.66

vous

-0.66

ttes

-0.66

 Rebels

-0.65

 Subtle

-0.65

POSITIVE LOGITS

ethyst

1.23

ulet

1.22

sterdam

1.21

endment

1.17

bitious

1.07

nesia

1.04

ajor

1.01

nesty

1.01

itan

1.00

onso

0.99

Activations Density 0.016%

occurrences of the term "Am" followed by a number

instances of the abbreviation "Am" followed by numbers, indicating references to specific entities or organizations

No Comments

No Known Activations