INDEX

Explanations

words related to email and system configuration settings, specifically focusing on actions related to managing emails and system settings

oai_token-act-pair · gpt-3.5-turbo

the word "your" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Originally

-0.82

trak

-0.76

 Uriel

-0.75

 Related

-0.73

 Strikes

-0.72

ilts

-0.71

 Seeking

-0.67

Pac

-0.67

-0.66

POSITIVE LOGITS

own

1.49

 opponent

1.04

 adversary

0.98

 favourite

0.98

 tremend

0.96

 favorite

0.93

ocard

0.92

 imagination

0.92

 respective

0.92

 desired

0.90

Activations Density 0.113%

words related to email and system configuration settings, specifically focusing on actions related to managing emails and system settings

the word "your" in various contexts

No Comments

No Known Activations