INDEX

Explanations

instances of the word "elect" and variations like "electric" or "electrical."

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

MENTS

-0.77

ĪĴ

-0.72

inka

-0.72

chest

-0.71

Ò

-0.66

ãĥĦ

-0.65

 Furious

-0.65

oku

-0.65

MENT

-0.64

sworth

-0.62

POSITIVE LOGITS

ounter

1.23

urrent

1.21

onduct

1.20

ount

1.20

ouple

1.19

ounters

1.19

urrency

1.18

redited

1.17

hester

1.16

itizen

1.15

Activations Density 0.718%

instances of the word "elect" and variations like "electric" or "electrical."

No Comments

No Known Activations