INDEX

Explanations

specific numerical information mixed with structured layout elements within texts, likely from documents or reports

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

forth

-0.83

alore

-0.68

ppard

-0.63

minist

-0.62

ilitarian

-0.60

 spont

-0.59

eur

-0.58

Corp

-0.58

Leaks

-0.56

AFTA

-0.55

POSITIVE LOGITS

 Featured

0.88

][

0.85

 Expand

0.80

 âĢº

0.76

 guiActiveUnfocused

0.76

actionDate

0.75

 SHARES

0.75

 Â·

0.74

 VIDEOS

0.72

0.71

Activations Density 1.214%

specific numerical information mixed with structured layout elements within texts, likely from documents or reports

No Comments

No Known Activations