INDEX

Explanations

phrases related to challenges, achievements, and the need to optimize performance

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

 Stranger

-0.61

 resides

-0.60

 belongs

-0.59

 Adolf

-0.59

âĺħ

-0.58

 Lover

-0.58

 revol

-0.57

 vanished

-0.57

 denotes

-0.56

 doubtless

-0.56

POSITIVE LOGITS

 ourselves

1.51

mble

0.94

)."

0.87

]."

0.84

catentry

0.78

 incent

0.76

.""

0.75

%"

0.75

ruce

0.73

 ours

0.73

Activations Density 0.489%

phrases related to challenges, achievements, and the need to optimize performance

No Comments

No Known Activations