INDEX

Explanations

mentions of rankings and positions in various contexts

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

 ranks

-0.69

 cutting

-0.67

 Britann

-0.66

 Xiao

-0.65

 Lawrence

-0.64

 Dickinson

-0.64

 Torch

-0.64

 Wiley

-0.63

 laure

-0.62

Tik

-0.62

POSITIVE LOGITS

distance

1.23

committee

1.19

advertising

1.17

speech

1.15

diff

1.15

prison

1.15

terms

1.13

function

1.13

recomm

1.11

cong

1.11

Activations Density 0.476%

mentions of rankings and positions in various contexts

No Comments

No Known Activations