INDEX

Explanations

phrases related to hindering or preventing actions

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

ocene

-0.80

rics

-0.74

Sov

-0.73

OLOGY

-0.71

iosyncr

-0.67

eah

-0.66

 Dynamics

-0.66

olate

-0.65

sburg

-0.65

nai

-0.64

POSITIVE LOGITS

gap

0.83

buster

0.81

 bothering

0.80

ions

0.76

 runaway

0.76

 blocking

0.76

 accessing

0.75

 breathing

0.74

 smoking

0.72

 extradition

0.71

Activations Density 1.969%

phrases related to hindering or preventing actions

No Comments

No Known Activations