INDEX

Explanations

statements or questions that speculate or analyze different perspectives or arguments on a topic

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

aukee

-0.79

andem

-0.72

Guard

-0.70

oses

-0.70

rack

-0.68

gur

-0.68

cellaneous

-0.65

atro

-0.63

mouth

-0.63

ivalent

-0.63

POSITIVE LOGITS

 although

1.01

 there

0.98

 they

0.97

 whoever

0.83

we

0.79

"[

0.78

soever

0.77

 THEY

0.75

 neither

0.75

 despite

0.74

Activations Density 3.022%

statements or questions that speculate or analyze different perspectives or arguments on a topic

No Comments

No Known Activations