INDEX

Explanations

phrases related to endings or conclusions, often denoted with specific symbols like âĢĶ and Ċ

oai_token-act-pair · gpt-3.5-turbo

strong emotional statements and expressions of hope or motivation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.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

 transport

-0.73

 colour

-0.73

 hardware

-0.73

 resent

-0.72

 armour

-0.71

 disg

-0.71

 capacity

-0.70

 offending

-0.70

 tentacles

-0.69

 quadru

-0.69

POSITIVE LOGITS

 Conclusion

1.11

 Volume

1.04

 Lessons

1.00

 Part

0.98

 Debate

0.95

 Revis

0.94

 Month

0.94

 Remastered

0.93

 Rankings

0.92

 Chapter

0.91

Activations Density 0.315%

phrases related to endings or conclusions, often denoted with specific symbols like âĢĶ and Ċ

strong emotional statements and expressions of hope or motivation

No Comments

No Known Activations