INDEX

Explanations

places or settings that are described in detail

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

ccording

-0.81

 lapse

-0.72

prus

-0.72

 turf

-0.69

å£«

-0.65

 extent

-0.65

 atmosphere

-0.65

 mosqu

-0.65

terday

-0.65

bryce

-0.64

POSITIVE LOGITS

erers

2.02

erer

1.91

ering

1.45

ered

1.18

ern

1.13

eful

0.99

ring

0.97

ers

0.95

eren

0.91

ishing

0.91

Activations Density 0.032%

places or settings that are described in detail

No Comments

No Known Activations