INDEX

Explanations

numeric or technical terms related to settings or configurations

oai_token-act-pair · gpt-3.5-turbo

instances of the word "parameters" and its variations in a technical or contextual setting

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.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

bery

-0.78

cott

-0.67

istle

-0.67

Tai

-0.67

atl

-0.66

itious

-0.64

 Rouge

-0.63

fur

-0.61

sold

-0.61

rice

-0.60

POSITIVE LOGITS

 parameters

3.66

 parameter

2.44

 Parameters

2.38

Parameters

1.97

 variables

1.94

Parameter

1.85

 params

1.77

param

1.71

params

1.66

 param

1.47

Activations Density 0.018%

numeric or technical terms related to settings or configurations

instances of the word "parameters" and its variations in a technical or contextual setting

No Comments

No Known Activations