INDEX

Explanations

places, countries, and other location-related terms

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

inventoryQuantity

-0.66

atin

-0.63

onym

-0.62

tops

-0.61

å§

-0.60

soDeliveryDate

-0.60

Discussion

-0.60

Sym

-0.60

HAEL

-0.59

iple

-0.59

POSITIVE LOGITS

 Airlines

1.08

 Arabia

0.91

's

0.85

 annexed

0.83

 Today

0.83

ESE

0.81

 invaded

0.81

 Yard

0.77

 Airways

0.77

 TODAY

0.75

Activations Density 1.411%

places, countries, and other location-related terms

No Comments

No Known Activations