INDEX

Explanations

legal and punitive terms

oai_token-act-pair · gpt-3.5-turbo

terms related to punitive actions and judgments

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

WAYS

-0.81

ords

-0.81

eds

-0.80

andestine

-0.80

ways

-0.80

roots

-0.80

 Fitz

-0.75

reads

-0.75

stones

-0.73

doms

-0.73

POSITIVE LOGITS

 punitive

1.30

 chast

0.99

rig

0.76

============

0.76

 crus

0.76

 pend

0.75

 punishing

0.73

 punishment

0.73

©¶æ

0.73

 damages

0.73

Activations Density 0.018%

legal and punitive terms

terms related to punitive actions and judgments

No Comments

No Known Activations