INDEX

Explanations

mentions of political news and controversies

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

ops

-0.72

viation

-0.69

ibo

-0.68

reth

-0.65

arily

-0.65

ulp

-0.65

ibe

-0.64

ogly

-0.64

ofer

-0.64

assed

-0.63

POSITIVE LOGITS

 rumors

0.96

 tensions

0.87

 rumours

0.80

 pundits

0.75

 however

0.71

 commentators

0.69

 speculation

0.68

 meanwhile

0.66

 Concern

0.65

 rumor

0.65

Activations Density 0.507%

mentions of political news and controversies

No Comments

No Known Activations