INDEX

Explanations

words related to instructions, guidelines, or recommendations

oai_token-act-pair · gpt-3.5-turbo

phrases emphasizing obligations or necessities

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

 Guilty

-0.67

pires

-0.66

 Lambert

-0.64

MAP

-0.64

mort

-0.62

Mur

-0.62

Ips

-0.62

lip

-0.61

MAP

-0.59

 Portable

-0.58

POSITIVE LOGITS

 rely

1.02

 abide

1.00

 wait

0.99

 obey

0.98

 worry

0.90

 prove

0.90

 contend

0.89

 comply

0.88

be

0.88

 remind

0.88

Activations Density 0.090%

words related to instructions, guidelines, or recommendations

phrases emphasizing obligations or necessities

No Comments

No Known Activations