INDEX

Explanations

phrases related to contrasting or introducing new information

oai_token-act-pair · gpt-3.5-turbo

conjunctions and coordinating phrases indicating connectivity or relationships between ideas

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

BuyableInstoreAndOnline

-0.76

tnc

-0.74

ãĥĦ

-0.73

matter

-0.73

inventoryQuantity

-0.72

factor

-0.71

dayName

-0.69

ibur

-0.68

Synopsis

-0.67

"]=>

-0.67

POSITIVE LOGITS

 many

1.45

 none

1.44

 others

1.35

 dozens

1.34

 some

1.33

few

1.28

many

1.25

 they

1.22

 countless

1.21

 several

1.21

Activations Density 0.506%

phrases related to contrasting or introducing new information

conjunctions and coordinating phrases indicating connectivity or relationships between ideas

No Comments

No Known Activations