INDEX

Explanations

mentions of things that happened or existed in the past

oai_token-act-pair · gpt-3.5-turbo

references to prior instances or events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

lee

-0.80

lua

-0.78

 ILCS

-0.74

hots

-0.72

rage

-0.68

kamp

-0.67

rosso

-0.66

 Franch

-0.65

cars

-0.65

aliation

-0.65

POSITIVE LOGITS

 generations

1.05

ebin

1.00

 incarnation

0.95

 incarn

0.90

 occupant

0.83

 batch

0.79

 iterations

0.78

 iteration

0.78

 installments

0.77

ĻĤ

0.74

Activations Density 0.019%

mentions of things that happened or existed in the past

references to prior instances or events

No Comments

No Known Activations