INDEX

Explanations

the word "good" and related words indicating positivity or high quality

oai_token-act-pair · gpt-3.5-turbo

phrases or contexts indicating positive quality or reassurance

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

hyde

-0.79

eters

-0.75

oths

-0.73

atum

-0.73

pper

-0.70

eds

-0.69

ategory

-0.67

 Tsukuyomi

-0.67

opers

-0.66

 Reincarnated

-0.65

POSITIVE LOGITS

enough

1.35

 luck

1.12

bye

1.05

reads

1.04

ol

1.03

luck

1.03

 enough

1.03

 Samar

1.02

 intentions

1.00

 quality

0.89

Activations Density 0.081%

the word "good" and related words indicating positivity or high quality

phrases or contexts indicating positive quality or reassurance

No Comments

No Known Activations