INDEX

Explanations

phrases related to prediction or possibility

oai_token-act-pair · gpt-3.5-turbo

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

romptu

-0.67

athing

-0.66

OHN

-0.65

ulas

-0.63

landers

-0.63

ICA

-0.63

Joined

-0.63

lash

-0.62

ENCY

-0.61

CRIP

-0.61

POSITIVE LOGITS

 hands

1.13

 realm

1.06

 forefront

0.96

 grasp

0.94

 vein

0.93

 minds

0.92

 drawer

0.88

 foreground

0.86

 pipeline

0.83

 paws

0.83

Activations Density 0.130%

phrases related to prediction or possibility

No Comments

No Known Activations