INDEX

Explanations

phrases that indicate reporting or news dissemination

oai_token-act-pair · gpt-4o-mini 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

izont

-0.85

ãĤ´ãĥ³

-0.83

ĸļ

-0.83

aho

-0.79

wow

-0.77

akable

-0.77

ãĤ¤ãĥĪ

-0.76

ocene

-0.75

ãĤ¦ãĤ¹

-0.74

ield

-0.71

POSITIVE LOGITS

 Sources

0.77

 sources

0.73

 rumors

0.73

 Pentagon

0.72

 eyewitness

0.72

 JPMorgan

0.71

 NYPD

0.71

 representatives

0.71

 rumor

0.70

FBI

0.69

Activations Density 0.221%

phrases that indicate reporting or news dissemination

No Comments

No Known Activations