INDEX

Explanations

discussion-related words and phrases

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

OGR

-0.95

 teasp

-0.74

externalActionCode

-0.73

cue

-0.71

©¶æ¥µ

-0.71

SS

-0.71

photo

-0.69

soever

-0.68

kie

-0.68

nis

-0.68

POSITIVE LOGITS

how

1.19

 whether

1.09

 topics

1.01

 reforming

0.96

 behalf

0.94

why

0.94

 improving

0.93

 issues

0.92

 preserving

0.89

 matters

0.86

Activations Density 4.108%

discussion-related words and phrases

No Comments

No Known Activations