INDEX

Explanations

references to "Che" or related culinary terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

CTR

-0.80

ãĥ¼ãĥĨãĤ£

-0.72

 executive

-0.66

 Icelandic

-0.64

ãĥ¤

-0.63

 Archdemon

-0.62

 Acts

-0.61

 EDITION

-0.60

inally

-0.59

 Atmospheric

-0.58

POSITIVE LOGITS

esy

1.08

vette

1.05

hou

1.04

cks

1.02

lean

1.00

apeake

0.98

bones

0.97

apest

0.96

ards

0.95

roo

0.95

Activations Density 0.012%

references to "Che" or related culinary terms

No Comments

No Known Activations