INDEX

Explanations

words related to alternatives and options in various contexts

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

frey

-0.80

eln

-0.78

rolling

-0.75

ooks

-0.72

hemat

-0.72

blogspot

-0.71

ucket

-0.69

ansk

-0.68

oho

-0.68

aping

-0.67

POSITIVE LOGITS

 thereof

0.98

 thereto

0.74

hesis

0.70

 mechanism

0.69

ãĥ¤

0.67

 list

0.66

imum

0.63

 Include

0.63

="#

0.63

kit

0.62

Activations Density 6.096%

words related to alternatives and options in various contexts

No Comments

No Known Activations