INDEX

Explanations

phrases related to coincidences or statistical correlations

oai_token-act-pair · gpt-3.5-turbo

references to incidents and occurrences

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

wo

-0.66

ign

-0.66

dri

-0.65

 Furn

-0.64

 unstoppable

-0.63

 Neutral

-0.63

 reactor

-0.63

Wad

-0.62

Dem

-0.62

 Payne

-0.62

POSITIVE LOGITS

idences

3.09

Previously

1.43

olson

1.32

 associations

1.26

akings

1.11

 Previously

1.08

ctors

1.05

ourses

1.02

ricks

1.02

 coli

1.00

Activations Density 0.051%

phrases related to coincidences or statistical correlations

references to incidents and occurrences

No Comments

No Known Activations