INDEX

Explanations

keywords associated with editing, modifying, and tweaking digital content

oai_token-act-pair · gpt-3.5-turbo

phrases related to modification and editing actions

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

SPONSORED

-0.75

thy

-0.74

sters

-0.72

tale

-0.71

 Adults

-0.70

nai

-0.69

AMI

-0.69

 Mothers

-0.69

 Senators

-0.67

 patrons

-0.64

POSITIVE LOGITS

 rearr

1.10

 resh

0.96

 decomp

0.95

 photoc

0.94

 delet

0.91

 reorgan

0.90

 reuse

0.90

 recomb

0.89

 rewriting

0.88

 subtract

0.87

Activations Density 0.284%

keywords associated with editing, modifying, and tweaking digital content

phrases related to modification and editing actions

No Comments

No Known Activations