INDEX

Explanations

phrases posing a question or issue

oai_token-act-pair · gpt-3.5-turbo

points of contention or debate in discussions

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

 supposedly

-0.70

 latter

-0.66

Ö¼

-0.64

 comparatively

-0.63

 ostensibly

-0.63

 largely

-0.63

 inexpl

-0.62

 purported

-0.61

 evidently

-0.61

 purportedly

-0.60

POSITIVE LOGITS

 Yourself

1.05

 yourself

1.02

 yourselves

0.89

 Your

0.84

 YOUR

0.83

your

0.80

 your

0.79

 Reduce

0.78

!:

0.77

 Submit

0.77

Activations Density 0.567%

phrases posing a question or issue

points of contention or debate in discussions

No Comments

No Known Activations