INDEX

Explanations

unrelated or contrasting statements within the same context

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

SourceFile

-0.85

arah

-0.70

ãĥ¥

-0.69

olves

-0.68

ULAR

-0.67

MpServer

-0.65

alysed

-0.65

arily

-0.64

ãĤ¼ãĤ¦ãĤ¹

-0.62

Cause

-0.62

POSITIVE LOGITS

 however

1.11

 there

0.97

 although

0.97

 though

0.89

 according

0.86

 unlike

0.86

 despite

0.86

we

0.84

 unless

0.83

 moreover

0.83

Activations Density 1.637%

unrelated or contrasting statements within the same context

No Comments

No Known Activations

unrelated or contrasting statements within the same context

No Comments

No Known Activations