INDEX

Explanations

specific phrases related to news headlines or video titles

oai_token-act-pair · gpt-3.5-turbo

instances of the phrase "JUST WATCHED."

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

sis

-0.74

sole

-0.69

ocr

-0.69

Magikarp

-0.67

ogly

-0.67

pill

-0.66

nib

-0.65

nuts

-0.65

nut

-0.65

 detached

-0.64

POSITIVE LOGITS

 WATCHED

1.37

 Transcript

0.81

 VIDEOS

0.81

 WATCH

0.73

 Countdown

0.73

 Videos

0.73

 Task

0.70

 Scenes

0.69

 IMAGES

0.69

 reacts

0.68

Activations Density 0.008%

specific phrases related to news headlines or video titles

instances of the phrase "JUST WATCHED."

No Comments

No Known Activations