INDEX

Explanations

repetitive prompts directed at the reader, emphasizing engagement and interaction

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

assic

-0.74

bush

-0.73

-0.71

ylum

-0.69

teen

-0.67

gement

-0.65

senal

-0.65

 Flags

-0.61

mining

-0.61

furt

-0.61

POSITIVE LOGITS

 afford

1.22

 imagine

1.12

 quantify

1.01

 tell

0.98

 convince

0.97

 please

0.95

 reconcile

0.95

 PLEASE

0.94

 conceive

0.92

 realistically

0.92

Activations Density 0.022%

repetitive prompts directed at the reader, emphasizing engagement and interaction

No Comments

No Known Activations