INDEX

Explanations

instances of direct speech actions like saying, asking, or telling

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

xtap

-0.88

cv

-0.83

wx

-0.75

osponsors

-0.74

irtual

-0.71

unal

-0.68

mania

-0.65

ugal

-0.65

=~=~

-0.64

£ı

-0.64

POSITIVE LOGITS

bye

1.15

 goodbye

1.14

 hello

1.09

 aloud

0.93

hi

0.83

hey

0.81

 nothing

0.80

 Amen

0.78

 loudly

0.77

yes

0.76

Activations Density 0.112%

instances of direct speech actions like saying, asking, or telling

No Comments

No Known Activations