Update RESEARCH_ARTICLE.md
Browse filesadded information about hardware
- RESEARCH_ARTICLE.md +10 -0
RESEARCH_ARTICLE.md
CHANGED
|
@@ -24,6 +24,16 @@ Only this final distilled version is included in the public release.
|
|
| 24 |
|
| 25 |
---
|
| 26 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
# 1. Motivation
|
| 28 |
|
| 29 |
Large generative models can possess all of the concepts necessary to satisfy a prompt while still failing to preserve the relationships between those concepts.
|
|
|
|
| 24 |
|
| 25 |
---
|
| 26 |
|
| 27 |
+
### Experimental Hardware
|
| 28 |
+
|
| 29 |
+
All experiments reported in this project were conducted locally using a single **NVIDIA GeForce RTX 3090 Ti (24 GB VRAM)**.
|
| 30 |
+
|
| 31 |
+
This includes hidden-state extraction from the source and target models, projector screening, semantic-bridge construction, student distillation, quantitative evaluation, and qualitative MiniMax H3 generation experiments.
|
| 32 |
+
|
| 33 |
+
No multi-GPU or datacenter training cluster was used.
|
| 34 |
+
|
| 35 |
+
This hardware constraint was intentional: part of the experiment was to determine whether cross-architecture semantic representation transfer could be explored and distilled into a practical adapter using consumer-accessible compute.
|
| 36 |
+
|
| 37 |
# 1. Motivation
|
| 38 |
|
| 39 |
Large generative models can possess all of the concepts necessary to satisfy a prompt while still failing to preserve the relationships between those concepts.
|