INDEX

Explanations

instances where the word "can" is used

oai_token-act-pair · gpt-3.5-turbo

contextual verbs that imply possibility or capability

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Fighter

-0.69

BAL

-0.63

Moz

-0.62

 honoring

-0.61

spection

-0.58

edient

-0.58

 Mant

-0.58

din

-0.58

 Generation

-0.58

 Quarter

-0.58

POSITIVE LOGITS

't

1.52

adian

1.24

berra

1.20

isters

1.09

NOT

1.03

ister

0.98

nery

0.93

 afford

0.92

 easily

0.92

vas

0.92

Activations Density 0.164%

instances where the word "can" is used

contextual verbs that imply possibility or capability

No Comments

No Known Activations