INDEX

Explanations

phrases related to public statements or opinions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "comments" and related expressions discussing remarks made by individuals

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

fruit

-0.73

 Bride

-0.69

PG

-0.68

 Recon

-0.68

ISH

-0.67

 Rescue

-0.67

BLIC

-0.66

ILD

-0.62

riz

-0.62

tom

-0.61

POSITIVE LOGITS

 remarks

0.94

 uttered

0.92

ariat

0.91

 comments

0.91

 dispar

0.75

 guiActiveUn

0.73

 slurs

0.73

 briefings

0.72

oras

0.72

 regarding

0.72

Activations Density 0.039%

phrases related to public statements or opinions

instances of the word "comments" and related expressions discussing remarks made by individuals

No Comments

No Known Activations