INDEX

Explanations

time-related adverbs or adverbial phrases

oai_token-act-pair · gpt-3.5-turbo

discourse markers indicating time

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

imm

-0.71

initely

-0.67

ajo

-0.60

nic

-0.60

neys

-0.57

ãĥ¯ãĥ³

-0.57

 Â®

-0.57

CHQ

-0.56

 Equality

-0.55

omination

-0.55

POSITIVE LOGITS

 thereafter

1.60

 afterwards

1.31

 afterward

1.28

 before

1.12

 beforehand

1.00

 after

1.00

ago

0.97

aft

0.93

 enough

0.91

before

0.90

Activations Density 0.024%

time-related adverbs or adverbial phrases

discourse markers indicating time

No Comments

No Known Activations