INDEX

Explanations

textual dialogue cues and storytelling elements, focusing on introspection and self-reflection

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

 guiActiveUnfocused

-0.86

 Palest

-0.82

Buyable

-0.67

 hors

-0.67

 scattering

-0.64

 Networks

-0.64

ozy

-0.63

 Izan

-0.62

RAD

-0.61

GPS

-0.61

POSITIVE LOGITS

£

1.01

should

0.97

¤

0.95

¼

0.94

Â¢

0.94

could

0.93

Ń

0.93

ª

0.91

º

0.90

Ĵ

0.89

Activations Density 0.659%

textual dialogue cues and storytelling elements, focusing on introspection and self-reflection

No Comments

No Known Activations