INDEX

Explanations

words related to political figures, particularly in situations involving investigation or controversy

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

reciation

-0.64

 ancest

-0.59

 hereafter

-0.59

 crunch

-0.59

 brace

-0.59

liest

-0.59

 dunno

-0.58

hold

-0.58

ngth

-0.57

cellent

-0.57

POSITIVE LOGITS

geist

0.73

Vi

0.70

lang

0.69

Cola

0.68

iform

0.67

é»Ĵ

0.65

ãĤ´ãĥ³

0.65

çĭ

0.65

Ped

0.65

Poly

0.65

Activations Density 0.077%

words related to political figures, particularly in situations involving investigation or controversy

No Comments

No Known Activations