INDEX

Explanations

specific names or terms, particularly those associated with literature and storytelling

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

ongyang

-0.85

 pestic

-0.69

 Playoffs

-0.67

 juices

-0.66

ulo

-0.63

 reps

-0.62

 circulation

-0.62

ecause

-0.60

 Featured

-0.60

 applic

-0.60

POSITIVE LOGITS

inki

1.10

furt

0.95

stad

0.87

beck

0.86

gow

0.85

kamp

0.80

ways

0.79

mann

0.79

forth

0.78

å§«

0.77

Activations Density 0.099%

specific names or terms, particularly those associated with literature and storytelling

No Comments

No Known Activations