INDEX

Explanations

mentions or predictions about future events

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

era

-0.65

buster

-0.61

plug

-0.58

wise

-0.58

>>\

-0.57

zer

-0.56

oot

-0.56

tip

-0.55

strip

-0.55

 endeavor

-0.55

POSITIVE LOGITS

 plenty

1.18

 ample

0.97

 lots

0.91

no

0.91

 some

0.86

 enough

0.84

 alot

0.83

 exceptions

0.83

 tons

0.82

ample

0.82

Activations Density 0.026%

mentions or predictions about future events

No Comments

No Known Activations