INDEX

Explanations

political terms and names

oai_token-act-pair · gpt-3.5-turbo

occurrences of punctuation, specifically commas

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

rem

-0.67

 output

-0.64

rid

-0.64

ith

-0.64

 concess

-0.62

manac

-0.62

 overview

-0.62

 outputs

-0.60

 capacity

-0.59

 experience

-0.58

POSITIVE LOGITS

Jr

0.92

aka

0.89

 Flavoring

0.84

 albeit

0.84

Sr

0.82

 although

0.82

Jr

0.80

 namely

0.79

 which

0.78

0.76

Activations Density 0.587%

political terms and names

occurrences of punctuation, specifically commas

No Comments

No Known Activations