INDEX

Explanations

locations, such as specific shops, streets, or gas stations

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

nir

-0.70

ãģĦ

-0.68

utan

-0.67

lust

-0.65

 Marketable

-0.65

thin

-0.65

CHAT

-0.65

 Rounds

-0.64

pas

-0.60

ï¸ı

-0.60

POSITIVE LOGITS

 located

0.97

 entrance

0.96

 premises

0.96

 entrances

0.94

keeper

0.91

 doors

0.90

 occupancy

0.89

 operated

0.88

keepers

0.88

house

0.86

Activations Density 2.676%

locations, such as specific shops, streets, or gas stations

No Comments

No Known Activations