Update README.md
Browse files
README.md
CHANGED
|
@@ -1,5 +1,9 @@
|
|
| 1 |
---
|
| 2 |
-
base_model:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
license: mit
|
| 4 |
datasets:
|
| 5 |
- arcee-ai/EvolKit-75K
|
|
@@ -14,12 +18,11 @@ Experimental commander model V1.
|
|
| 14 |
|
| 15 |
Named it Zelensky in order to troll Uncle Elon on twitter over how bad Grok-2 is.
|
| 16 |
|
| 17 |
-
Training process, low 1 epoch learning rate and evolutionary-merged via https://github.com/arcee-ai/EvolKit
|
| 18 |
|
| 19 |
-
Process on 8x AMD Mi300 192GB gpus.
|
| 20 |
|
| 21 |
Thank you Vultr https://www.vultr.com/register/ for sponsoring the compute.
|
| 22 |
|
| 23 |
|
| 24 |
-
|
| 25 |
-
Qwen License applies by default.
|
|
|
|
| 1 |
---
|
| 2 |
+
base_model:
|
| 3 |
+
- Qwen/Qwen2.5-72B-Instruct
|
| 4 |
+
- huihui-ai/Qwen2.5-72B-Instruct-abliterated
|
| 5 |
+
- Qwen/Qwen2.5-72B
|
| 6 |
+
- spow12/ChatWaifu_72B_v2.2
|
| 7 |
license: mit
|
| 8 |
datasets:
|
| 9 |
- arcee-ai/EvolKit-75K
|
|
|
|
| 18 |
|
| 19 |
Named it Zelensky in order to troll Uncle Elon on twitter over how bad Grok-2 is.
|
| 20 |
|
| 21 |
+
Training process, low 1 epoch learning rate and evolutionary-merged with the 3 other listed models via https://github.com/arcee-ai/EvolKit
|
| 22 |
|
| 23 |
+
Process repeated multiple times on 8x AMD Mi300 192GB gpus while also running gpqa_diamond_zeroshot on LM_Eval harness.
|
| 24 |
|
| 25 |
Thank you Vultr https://www.vultr.com/register/ for sponsoring the compute.
|
| 26 |
|
| 27 |
|
| 28 |
+
Qwen License still applies by default.
|
|
|