INDEX

Explanations

dates and numerical values within a context

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

awaru

-0.75

 Lumpur

-0.70

orative

-0.65

 mascul

-0.62

 hygiene

-0.57

 propri

-0.57

espie

-0.56

 Pompe

-0.55

ategory

-0.55

 hook

-0.55

POSITIVE LOGITS

st

1.43

Password

1.33

1.30

1.19

1.13

Â½

1.10

1.03

âģ

1.02

0.99

Activations Density 0.958%

dates and numerical values within a context

No Comments

No Known Activations