INDEX

Explanations

words related to criticism and evaluation

oai_token-act-pair · gpt-3.5-turbo

phrases and words indicating performance or quality evaluations, often with a focus on adherence to standards or criteria

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 cost

-0.38

 costs

-0.38

 Daytona

-0.37

enegger

-0.36

 Beir

-0.36

isSpecialOrderable

-0.36

 costing

-0.35

 affiliate

-0.35

 acquired

-0.35

 Dominican

-0.34

POSITIVE LOGITS

etheless

0.76

NetMessage

0.75

terness

0.60

auc

0.49

cing

0.48

aging

0.47

theless

0.46

amb

0.46

imb

0.46

ouncing

0.45

Activations Density 2.847%

words related to criticism and evaluation

phrases and words indicating performance or quality evaluations, often with a focus on adherence to standards or criteria

No Comments

No Known Activations