INDEX

Explanations

action words in the form of verbs that relate to accomplishing or achieving something

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate the ability to achieve or accomplish something

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

grounds

-0.78

 Perspect

-0.76

 Directions

-0.75

 Respons

-0.75

quartered

-0.74

listed

-0.71

mails

-0.71

 Principles

-0.70

 Integrity

-0.70

plates

-0.69

POSITIVE LOGITS

 overcome

1.15

 accomplish

1.14

 convince

1.10

 penetrate

1.07

 achieve

1.06

 solve

1.03

 evade

1.03

get

1.01

 raise

1.01

 recreate

1.00

Activations Density 0.086%

action words in the form of verbs that relate to accomplishing or achieving something

phrases that indicate the ability to achieve or accomplish something

No Comments

No Known Activations