INDEX

Explanations

forms of the verb "to be."

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

 totality

-0.80

quo

-0.77

 tens

-0.74

 societies

-0.73

 acknow

-0.73

 directions

-0.72

 situations

-0.72

 dart

-0.71

 areas

-0.71

 flare

-0.71

POSITIVE LOGITS

 Wrong

1.05

 Them

1.02

Us

0.99

 Stupid

0.99

 Killed

0.98

 Possible

0.98

 Hate

0.98

Own

0.97

 Changed

0.96

 Hits

0.93

Activations Density 0.057%

forms of the verb "to be."

No Comments

No Known Activations