INDEX

Explanations

messages addressed to someone specific

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_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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

bill

-0.74

atsu

-0.69

FN

-0.68

igree

-0.67

 Canad

-0.67

MH

-0.66

atem

-0.66

ucle

-0.66

 largeDownload

-0.65

tions

-0.65

POSITIVE LOGITS

 readers

1.18

 aspiring

1.09

 policymakers

1.00

 listeners

0.98

 youngsters

0.88

 beginners

0.85

 drinkers

0.83

 voters

0.83

 Readers

0.82

 believers

0.81

Activations Density 0.200%

messages addressed to someone specific

No Comments

No Known Activations