INDEX

Explanations

phrases related to technical instructions or coding

oai_token-act-pair · gpt-3.5-turbo

technical terms related to processes and systems

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

IQ

-0.69

 Bild

-0.67

oni

-0.66

ashtra

-0.66

 biases

-0.64

 undercover

-0.64

ague

-0.61

stakes

-0.60

ahi

-0.60

ason

-0.60

POSITIVE LOGITS

 restoration

1.78

 restored

1.76

 restore

1.63

 restoring

1.62

 revive

1.62

 restores

1.60

 revival

1.53

 rejuven

1.53

 revived

1.52

 rebuilt

1.51

Activations Density 0.785%

phrases related to technical instructions or coding

technical terms related to processes and systems

No Comments

No Known Activations