INDEX

Explanations

high-level disappointment or disapproval in reviews or comments

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

 enrol

-0.87

 authorized

-0.87

 credential

-0.83

 waged

-0.77

 emergency

-0.73

 authorised

-0.73

 invitation

-0.73

 inund

-0.73

 unmanned

-0.73

idal

-0.72

POSITIVE LOGITS

 Anyway

1.92

 Especially

1.67

Anyway

1.63

 Regardless

1.62

 Luckily

1.57

 Hopefully

1.56

 However

1.55

 Otherwise

1.54

 Also

1.53

 Thankfully

1.53

Activations Density 1.742%

high-level disappointment or disapproval in reviews or comments

No Comments

No Known Activations