INDEX

Explanations

phrases or sentences that contain contrasting elements or ideas

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

 Travels

-0.72

 activ

-0.61

 culminated

-0.57

 conven

-0.57

 culmin

-0.57

aw

-0.56

 personalized

-0.56

 redes

-0.55

 accelerated

-0.55

ize

-0.54

POSITIVE LOGITS

nor

1.89

nor

1.60

Nor

1.56

Nor

1.31

 Instead

1.30

yet

1.25

Instead

1.23

 Neither

1.19

unless

1.16

 neither

1.16

Activations Density 3.393%

phrases or sentences that contain contrasting elements or ideas

No Comments

No Known Activations