INDEX

Explanations

positive descriptors or attributes associated with experiences, people, or objects

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

ogens

-0.82

eligible

-0.72

arians

-0.72

 Administ

-0.71

sic

-0.71

aer

-0.70

orig

-0.69

igate

-0.69

igating

-0.68

omics

-0.68

POSITIVE LOGITS

 breeze

0.91

 nice

0.85

 little

0.85

 additions

0.81

 bonus

0.80

 touches

0.80

 touch

0.79

bye

0.79

 neat

0.78

 gesture

0.78

Activations Density 0.018%

positive descriptors or attributes associated with experiences, people, or objects

No Comments

No Known Activations