INDEX

Explanations

phrases related to legal situations or courtroom drama

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

reddits

-0.77

incial

-0.72

iture

-0.69

_-

-0.69

ilib

-0.67

multi

-0.65

ignt

-0.64

 Cosponsors

-0.62

racuse

-0.62

itures

-0.61

POSITIVE LOGITS

 nowhere

1.11

wed

0.93

 necessity

0.89

 desperation

0.84

 curiosity

0.83

 bounds

0.83

 sheer

0.81

 boredom

0.80

 frustration

0.80

 thin

0.72

Activations Density 0.058%

phrases related to legal situations or courtroom drama

No Comments

No Known Activations