INDEX

Explanations

adjectives related to emotional states

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Baron

-0.71

 Fine

-0.65

 Firm

-0.62

 mediation

-0.61

 Shell

-0.60

 arbitration

-0.58

 Mund

-0.57

 Brig

-0.57

Uni

-0.57

 Christy

-0.57

POSITIVE LOGITS

usual

0.84

":[

0.83

enth

0.79

ever

0.77

pees

0.77

fters

0.76

ago

0.76

vous

0.74

 predecessors

0.71

ilee

0.70

Activations Density 0.165%

adjectives related to emotional states

No Comments

No Known Activations