INDEX

Explanations

phrases related to legal agreements and commitments

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 spotting

-0.76

 Genie

-0.70

 brainstorm

-0.67

 Revival

-0.66

 Chimera

-0.65

COMPLE

-0.65

ovember

-0.64

ilet

-0.64

 Makoto

-0.64

history

-0.64

POSITIVE LOGITS

 belonged

1.00

 belongs

0.95

 belong

0.93

 accustomed

0.90

 allegiance

0.90

 subscrib

0.88

 subscribe

0.88

 entitle

0.87

 subscribed

0.85

 attached

0.85

Activations Density 0.073%

phrases related to legal agreements and commitments

No Comments

No Known Activations