INDEX

Explanations

instances where an action or expectation is contrasted with an unexpected outcome beginning with "Instead."

oai_token-act-pair · gpt-3.5-turbo

phrases indicating contrast or alternatives

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

uble

-0.74

 identifiable

-0.66

 describ

-0.66

 categorized

-0.65

 Likes

-0.65

inguishable

-0.64

 classify

-0.64

 Intellectual

-0.64

acea

-0.62

 copyright

-0.61

POSITIVE LOGITS

Instead

1.15

 Nope

1.10

 alas

1.08

Wr

1.07

 Instead

1.07

 instead

1.00

instead

0.88

 Turns

0.86

 disappoint

0.74

 transpired

0.73

Activations Density 0.391%

instances where an action or expectation is contrasted with an unexpected outcome beginning with "Instead."

phrases indicating contrast or alternatives

No Comments

No Known Activations