INDEX

Explanations

phrases related to length or duration

oai_token-act-pair · gpt-3.5-turbo

references to the length and format of media presentations versus condensed versions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

IELD

-0.76

apego

-0.76

wcsstore

-0.75

 coerc

-0.71

uyomi

-0.71

 ambassadors

-0.70

¬¼

-0.69

soType

-0.67

INC

-0.65

 Pigs

-0.63

POSITIVE LOGITS

 chapters

1.33

 exposition

1.13

chapter

1.13

contained

1.11

length

1.11

 length

1.10

 Chapters

1.07

 lengthy

1.07

 narrated

1.05

 narration

1.03

Activations Density 0.468%

phrases related to length or duration

references to the length and format of media presentations versus condensed versions

No Comments

No Known Activations