INDEX

Explanations

references to specific movies or franchises

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

mouth

-0.87

mel

-0.79

onga

-0.72

strate

-0.65

inker

-0.65

inges

-0.64

phys

-0.64

gling

-0.64

istics

-0.64

ogly

-0.64

POSITIVE LOGITS

 Awakens

0.74

ÃįÃį

0.73

âĦ¢:

0.72

hler

0.70

 Episode

0.70

 Skywalker

0.70

itialized

0.68

 Galaxy

0.66

film

0.65

 galaxy

0.64

Activations Density 0.010%

references to specific movies or franchises

No Comments

No Known Activations