INDEX

Explanations

text related to physical instructions or steps

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

!.

-0.77

%.

-0.69

$.

-0.66

,...

-0.65

+.

-0.64

*.

-0.63

 although

-0.63

'.

-0.63

';

-0.61

HY

-0.61

POSITIVE LOGITS

pires

0.89

 depends

0.73

 constitutes

0.72

 entails

0.69

pired

0.69

 involves

0.68

isn

0.66

 mattered

0.63

 varies

0.62

 implies

0.62

Activations Density 3.163%

text related to physical instructions or steps

No Comments

No Known Activations