INDEX

Explanations

names of cities and sports teams

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

uador

-0.91

merce

-0.82

awaru

-0.75

orters

-0.74

loading

-0.74

vana

-0.73

rieve

-0.72

ocene

-0.70

olkien

-0.70

ktop

-0.70

POSITIVE LOGITS

 Celtics

1.10

 Harbor

1.10

 Bruins

1.07

 Dynamics

1.06

 Marathon

1.03

 Globe

1.03

 Garden

0.97

0.93

 College

0.90

 Consulting

0.89

Activations Density 0.021%

names of cities and sports teams

No Comments

No Known Activations