INDEX

Explanations

text related to political events and figures

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

 scourge

-0.76

 haul

-0.75

 instinct

-0.72

 hust

-0.71

 bounded

-0.70

 reven

-0.70

 piping

-0.68

 orche

-0.68

plet

-0.67

 intended

-0.67

POSITIVE LOGITS

 Lastly

1.77

 Finally

1.64

 Anyway

1.51

 Nevertheless

1.48

 Nonetheless

1.47

 Meanwhile

1.46

 Additionally

1.45

 Eventually

1.44

 Moreover

1.43

 Regardless

1.43

Activations Density 1.847%

text related to political events and figures

No Comments

No Known Activations