INDEX

Explanations

instructions or commands within a text

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "instruction" and its related forms

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

wolves

-0.80

oney

-0.75

bnb

-0.71

pes

-0.67

 Photographer

-0.64

perties

-0.64

wart

-0.62

Clo

-0.62

peat

-0.62

rencies

-0.61

POSITIVE LOGITS

 instruction

3.89

 Instruction

2.98

 instructions

2.36

 instructional

2.12

 Instructions

1.66

Instruct

1.62

 Instruct

1.56

ruction

1.55

 instruct

1.48

 instructor

1.39

Activations Density 0.026%

instructions or commands within a text

occurrences of the word "instruction" and its related forms

No Comments

No Known Activations