INDEX

Explanations

proper names and titles, particularly related to people and their roles

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

yip

-0.77

ulhu

-0.70

ById

-0.67

Premium

-0.66

ishable

-0.63

nces

-0.62

vy

-0.62

oyer

-0.61

monds

-0.61

OTE

-0.60

POSITIVE LOGITS

Dud

0.74

Dos

0.73

 Gaal

0.71

 Garcia

0.69

 Wilhelm

0.68

 Luigi

0.68

 Maced

0.66

 Maria

0.65

 Sawyer

0.65

 Ferdinand

0.65

Activations Density 0.034%

proper names and titles, particularly related to people and their roles

No Comments

No Known Activations