INDEX

Explanations

the word "finale" in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to television or series finales

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

gun

-0.75

holding

-0.71

guns

-0.71

olds

-0.70

ald

-0.70

urg

-0.69

ums

-0.69

articles

-0.69

states

-0.69

lying

-0.68

POSITIVE LOGITS

 finale

1.23

adelphia

0.91

 adolesc

0.90

 prematurely

0.82

 conclud

0.82

[|

0.81

ÃįÃį

0.80

 showdown

0.79

 spoiler

0.79

 reel

0.79

Activations Density 0.006%

the word "finale" in various contexts

references to television or series finales

No Comments

No Known Activations