INDEX

Explanations

phrases related to political figures and the White House

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

 ILCS

-0.81

amaz

-0.75

#$#$

-0.71

NetMessage

-0.71

REG

-0.71

ashtra

-0.70

iser

-0.68

Ô

-0.67

erala

-0.67

juries

-0.64

POSITIVE LOGITS

 aides

1.03

 Correspond

1.02

 Counsel

0.98

 Office

0.95

 counsel

0.94

 aide

0.94

 staffers

0.90

 adviser

0.89

 advisers

0.88

 Situation

0.87

Activations Density 0.760%

phrases related to political figures and the White House

No Comments

No Known Activations