INDEX

Explanations

tokens or characters that represent non-English or unusual text formats

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

iaries

-1.12

iary

-0.88

acles

-0.79

iance

-0.77

vana

-0.77

gling

-0.77

ippers

-0.77

onomy

-0.74

espie

-0.73

essa

-0.73

POSITIVE LOGITS

Ð¼

1.20

Ðº

1.17

Ðµ

1.17

Ð»

1.13

ÑĤ

1.07

Ð

1.06

Ñ

1.04

Ñģ

0.99

Ð²

0.97

Ð½

0.90

Activations Density 0.003%

tokens or characters that represent non-English or unusual text formats

No Comments

No Known Activations