INDEX

Explanations

phrases related to describing relationships, especially in terms of comparison and contrast

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

OGR

-0.87

unc

-0.83

gow

-0.81

vous

-0.73

nell

-0.72

TPPStreamerBot

-0.71

nit

-0.71

bye

-0.71

quished

-0.67

kamp

-0.67

POSITIVE LOGITS

 sexes

1.21

 genders

1.07

 halves

0.97

 extremes

0.86

 sides

0.84

 factions

0.81

 combatants

0.80

 eras

0.76

two

0.75

 conflicting

0.71

Activations Density 0.857%

phrases related to describing relationships, especially in terms of comparison and contrast

No Comments

No Known Activations