INDEX

Explanations

instances of the word "ple" or its variations, indicating a focus on pleasure-related concepts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

ERA

-0.77

 Kenobi

-0.76

LER

-0.75

uador

-0.73

quickShipAvailable

-0.72

~~~~

-0.70

NRS

-0.70

lar

-0.69

artifacts

-0.68

Versions

-0.68

POSITIVE LOGITS

teness

1.33

asure

0.97

thora

0.97

mented

0.95

xt

0.86

ASE

0.84

ple

0.82

asures

0.82

ments

0.81

ased

0.81

Activations Density 0.006%

instances of the word "ple" or its variations, indicating a focus on pleasure-related concepts

No Comments

No Known Activations