INDEX

Explanations

numerals

oai_token-act-pair · gpt-3.5-turbo

significant numerical values or references to measurements

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

âĺħâĺħ

-0.76

ADRA

-0.75

ãĤŃ

-0.74

ãģĵ

-0.74

éĥ

-0.72

åī

-0.71

ãĤ¹ãĥĪ

-0.70

 Jagu

-0.70

 Helm

-0.70

ãĤ¶

-0.69

POSITIVE LOGITS

erent

1.05

ividual

0.98

bably

0.97

lished

0.95

cture

0.92

ween

0.92

ledged

0.91

ng

0.88

ittle

0.87

ted

0.87

Activations Density 0.569%

numerals

significant numerical values or references to measurements

No Comments

No Known Activations