INDEX

Explanations

measurements or objects that undergo measuring

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "measuring" and related measurement terms

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

 Exile

-0.75

 Syndicate

-0.69

aren

-0.62

join

-0.61

leck

-0.61

icia

-0.60

 Reno

-0.60

ka

-0.59

 Synd

-0.59

 Brotherhood

-0.58

POSITIVE LOGITS

 measuring

3.67

 measurement

2.40

asuring

2.16

Measure

2.03

 measurements

2.03

 measured

1.85

 Measure

1.82

 measure

1.80

 estimating

1.69

 assessing

1.59

Activations Density 0.019%

measurements or objects that undergo measuring

occurrences of the word "measuring" and related measurement terms

No Comments

No Known Activations