INDEX

Explanations

numbers, units of measurement, and penalties

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

ople

-0.82

 Machines

-0.80

OPLE

-0.77

 Clockwork

-0.75

Mach

-0.73

bda

-0.72

olin

-0.72

Leon

-0.71

Machine

-0.71

ocial

-0.71

POSITIVE LOGITS

 recommendation

1.17

 implication

1.09

 suggestion

1.08

 requirement

1.05

 clue

1.02

 clause

1.02

 hint

0.97

 indication

0.96

 remark

0.96

 reservation

0.95

Activations Density 0.313%

numbers, units of measurement, and penalties

No Comments

No Known Activations