INDEX

Explanations

phrases related to using something for a specific purpose or action

oai_token-act-pair · gpt-3.5-turbo

the word "to" used in various 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

 Needs

-0.73

soDeliveryDate

-0.71

 Requirements

-0.64

 owes

-0.62

 Caller

-0.61

 lines

-0.60

 warranties

-0.60

 consequences

-0.59

icipated

-0.58

 Bhar

-0.58

POSITIVE LOGITS

 maximize

1.10

 promote

1.10

 create

1.04

 avoid

1.03

 facilitate

1.03

 enhance

1.02

 ensure

1.01

 lure

1.00

 preserve

1.00

 recreate

0.99

Activations Density 0.174%

phrases related to using something for a specific purpose or action

the word "to" used in various contexts

No Comments

No Known Activations