INDEX

Explanations

actions related to work or effort

oai_token-act-pair · gpt-3.5-turbo

phrases indicating efforts or actions taken towards a goal or outcome

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

 glances

-0.70

 Used

-0.70

quartered

-0.69

 Cust

-0.68

faces

-0.67

 Malf

-0.66

 curs

-0.65

 Spaces

-0.65

 Moments

-0.64

Lot

-0.62

POSITIVE LOGITS

 promote

1.28

ggles

1.20

 improve

1.19

 minimize

1.15

 emulate

1.15

 conserve

1.14

 achieve

1.12

 accommodate

1.12

 preserve

1.11

 ensure

1.11

Activations Density 0.203%

actions related to work or effort

phrases indicating efforts or actions taken towards a goal or outcome

No Comments

No Known Activations