INDEX

Explanations

phrases indicating a large quantity or proportion

oai_token-act-pair · gpt-3.5-turbo

phrases emphasizing the frequency and magnitude of 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

Ec

-0.66

î

-0.64

esc

-0.63

ÃĽ

-0.62

Sy

-0.61

¶æ

-0.60

 persu

-0.60

 forwards

-0.59

ety

-0.59

eteen

-0.59

POSITIVE LOGITS

LOT

3.04

 disproportionate

1.90

lot

1.56

 disproportionately

1.52

lot

1.35

Lot

1.14

Lot

1.11

 tremendous

1.09

little

1.07

 alot

1.06

Activations Density 0.016%

phrases indicating a large quantity or proportion

phrases emphasizing the frequency and magnitude of occurrences

No Comments

No Known Activations