INDEX

Explanations

phrases indicating a key component or principle in various contexts

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

DERR

-0.86

DragonMagazine

-0.82

scrib

-0.77

 Cosponsors

-0.77

FML

-0.71

sbm

-0.70

scl

-0.70

letter

-0.69

fif

-0.67

visory

-0.64

POSITIVE LOGITS

 unlocking

1.39

 success

1.15

 solving

1.11

 understanding

1.08

 achieving

1.04

 determining

1.03

 overcoming

1.02

 victory

0.98

 salvation

0.98

 restoring

0.95

Activations Density 0.046%

phrases indicating a key component or principle in various contexts

No Comments

No Known Activations