INDEX

Explanations

proper nouns and dates in news-like articles

oai_token-act-pair · gpt-3.5-turbo

punctuation marks and indicators of breaking news or updates

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

 endeavour

-0.84

lling

-0.75

dra

-0.72

 glide

-0.71

 teleport

-0.71

 abandon

-0.70

 apprentices

-0.69

 equip

-0.69

 Aval

-0.68

 borrow

-0.67

POSITIVE LOGITS

SPONSORED

1.36

Anonymous

1.21

According

1.08

Examples

1.07

Interestingly

1.07

However

1.07

Anyway

1.05

Apparently

1.05

Also

1.04

Furthermore

1.03

Activations Density 0.966%

proper nouns and dates in news-like articles

punctuation marks and indicators of breaking news or updates

No Comments

No Known Activations