INDEX

Explanations

specific patterns of numbers or phrases containing numbers

oai_token-act-pair · gpt-3.5-turbo

numerical references and series in the text

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

Southern

-0.61

 wedd

-0.57

 reproductive

-0.56

 sher

-0.55

 purse

-0.54

eral

-0.54

crim

-0.54

wr

-0.53

isch

-0.53

ivered

-0.53

POSITIVE LOGITS

1.16

1.12

1.10

1.09

1.07

1.06

1.02

1.00

0.99

0.97

Activations Density 0.058%

specific patterns of numbers or phrases containing numbers

numerical references and series in the text

No Comments

No Known Activations