INDEX

Explanations

words related to changes or updates

oai_token-act-pair · gpt-3.5-turbo

negative prefixes and terms associated with prohibition or limitation

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

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

 EntityItem

-0.95

 Samp

-0.77

 Cohn

-0.77

 Ross

-0.76

 Tune

-0.76

CHO

-0.76

 Shade

-0.76

 stare

-0.75

 shade

-0.74

IQ

-0.73

POSITIVE LOGITS

famous

1.54

dead

1.49

successful

1.45

popular

1.44

existing

1.44

empty

1.41

recent

1.40

exclusive

1.39

failed

1.37

prev

1.37

Activations Density 0.034%

words related to changes or updates

negative prefixes and terms associated with prohibition or limitation

No Comments

No Known Activations