INDEX

Explanations

mentions of time periods, specifically years

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "year."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

/_

-0.63

adders

-0.62

ignt

-0.61

alth

-0.61

creen

-0.60

Null

-0.58

galitarian

-0.57

Done

-0.55

krit

-0.54

aku

-0.53

POSITIVE LOGITS

 year

3.33

 Year

2.40

Year

2.39

 YEAR

2.35

 month

2.32

year

2.31

decade

2.10

 week

1.97

 season

1.84

yr

1.77

Activations Density 0.082%

mentions of time periods, specifically years

references to the concept of "year."

No Comments

No Known Activations