INDEX

Explanations

references to political and social hierarchies

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

".

-0.75

+.

-0.74

'.

-0.69

$.

-0.68

".

-0.65

!".

-0.65

"!

-0.63

."

-0.60

";

-0.60

?".

-0.59

POSITIVE LOGITS

pires

0.72

 allied

0.66

pired

0.63

 others

0.63

 countless

0.60

 successors

0.59

 assorted

0.58

 cohorts

0.56

 commentators

0.56

 venerable

0.56

Activations Density 0.233%

references to political and social hierarchies

No Comments

No Known Activations