INDEX

Explanations

specific phrasing indicating a statement or claim being made

oai_token-act-pair · gpt-3.5-turbo

claims or propositions that introduce a concept or idea requiring validation

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

ahime

-0.74

ugu

-0.70

onday

-0.68

————

-0.65

tu

-0.64

iola

-0.63

ached

-0.62

cms

-0.61

ikuman

-0.60

insert

-0.60

POSITIVE LOGITS

 horr

1.11

 haun

1.00

 hasn

1.00

 begs

0.97

 occurs

0.97

 bothers

0.97

 deserves

0.96

 distinguishes

0.95

 persists

0.95

 seems

0.94

Activations Density 0.151%

specific phrasing indicating a statement or claim being made

claims or propositions that introduce a concept or idea requiring validation

No Comments

No Known Activations