INDEX

Explanations

phrases indicating something will be addressed or discussed later

oai_token-act-pair · gpt-3.5-turbo

phrases indicating postponement or future discussion

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.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

Pwr

-0.83

Merit

-0.72

owship

-0.70

Ĥª

-0.65

HY

-0.64

 Haram

-0.62

washing

-0.61

Pri

-0.61

Ble

-0.61

mad

-0.61

POSITIVE LOGITS

ally

0.92

 generations

0.89

noon

0.86

 phases

0.84

aneously

0.83

 batches

0.81

 stages

0.79

 iterations

0.76

 installments

0.75

ality

0.74

Activations Density 0.032%

phrases indicating something will be addressed or discussed later

phrases indicating postponement or future discussion

No Comments

No Known Activations