INDEX

Explanations

references to media sources and publications

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

viation

-0.77

 disbanded

-0.71

Ã¨re

-0.66

ector

-0.64

orically

-0.64

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.64

hement

-0.63

 initials

-0.62

ãĤ±

-0.61

unda

-0.61

POSITIVE LOGITS

 Coverage

0.88

<|endoftext|>

0.87

 Stories

0.83

 Photos

0.81

 roundup

0.75

 Related

0.74

 Recent

0.73

Gad

0.73

 Helpful

0.73

Topics

0.72

Activations Density 0.067%

references to media sources and publications

No Comments

No Known Activations