INDEX

Explanations

numbers and units of measurement

oai_token-act-pair · gpt-3.5-turbo

numerical data and percentages related to economic or statistical trends

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 prosec

-0.90

 sculpt

-0.87

 reneg

-0.84

 oath

-0.81

 incarn

-0.79

 shroud

-0.79

 accus

-0.78

 lips

-0.78

 agon

-0.78

 amph

-0.77

POSITIVE LOGITS

Average

1.63

Percent

1.59

Overall

1.57

Compared

1.48

Among

1.43

Statistics

1.43

Numbers

1.40

Chart

1.40

Ranked

1.38

Looking

1.33

Activations Density 0.488%

numbers and units of measurement

numerical data and percentages related to economic or statistical trends

No Comments

No Known Activations