INDEX

Explanations

words related to geopolitical and political issues

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

MER

-0.71

Rye

-0.64

 Glob

-0.64

 saturation

-0.62

 Habit

-0.62

emin

-0.62

retion

-0.61

 Globe

-0.61

 drying

-0.61

GV

-0.60

POSITIVE LOGITS

soever

1.18

oping

1.08

 cares

1.07

ever

1.03

abouts

0.99

 knows

0.98

 else

0.98

 cared

0.92

oped

0.91

 owns

0.91

Activations Density 0.323%

words related to geopolitical and political issues

No Comments

No Known Activations