INDEX

Explanations

ellipses or pauses in text often represented by sequences of dots

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

ãĥ¼ãĥĨãĤ£

-0.82

isers

-0.70

 nutrit

-0.69

ocratic

-0.69

 arrow

-0.66

 fict

-0.66

 cape

-0.66

ãĥ¼ãĥĨ

-0.65

 roofs

-0.65

 credential

-0.64

POSITIVE LOGITS

etc

0.94

there

0.94

mmmm

0.92

THIS

0.85

sit

0.85

hammad

0.84

rss

0.83

interesting

0.83

well

0.83

these

0.83

Activations Density 0.005%

ellipses or pauses in text often represented by sequences of dots

No Comments

No Known Activations