INDEX

Explanations

dates in a specific format

oai_token-act-pair · gpt-3.5-turbo

date references in a specific format

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 accus

-0.72

izon

-0.68

eston

-0.67

onym

-0.65

 antiv

-0.64

Pow

-0.61

lain

-0.61

CLUD

-0.61

otine

-0.59

 wand

-0.58

POSITIVE LOGITS

å¹

0.88

0.85

0.83

January

0.81

0.80

released

0.79

release

0.78

February

0.78

Activations Density 0.095%

dates in a specific format

date references in a specific format

No Comments

No Known Activations