INDEX

Explanations

phrases related to making strong, often false statements

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

omb

-0.70

combe

-0.69

arrow

-0.68

schild

-0.68

patch

-0.67

izo

-0.65

å¸

-0.64

 Selected

-0.64

 Closed

-0.63

bern

-0.62

POSITIVE LOGITS

 innocence

0.94

 falsely

0.85

edly

0.84

 asylum

0.75

WTC

0.72

 ownership

0.71

 ignorance

0.68

 respons

0.68

 responsibility

0.66

 superiority

0.65

Activations Density 0.636%

phrases related to making strong, often false statements

No Comments

No Known Activations