INDEX

Explanations

phrases related to news and information

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

rom

-0.81

roph

-0.78

wal

-0.76

=-=-=-=-=-=-=-=-

-0.74

rosso

-0.73

Adv

-0.72

rolling

-0.72

rod

-0.71

rology

-0.69

gerald

-0.68

POSITIVE LOGITS

ity

1.10

ITY

0.97

 conception

0.89

iator

0.87

 impetus

0.86

 premise

0.83

 incarnation

0.81

 trilogy

0.81

 intent

0.79

izations

0.78

Activations Density 0.430%

phrases related to news and information

No Comments

No Known Activations