Upload merged model (patched README metadata)
Browse files
README.md
CHANGED
|
@@ -38,9 +38,9 @@ tool use, and recovery from errors.
|
|
| 38 |
- Epochs: 3
|
| 39 |
- Learning rate: 8e-06
|
| 40 |
- LoRA: r=64, alpha=128
|
| 41 |
-
- warmup_ratio : 0
|
| 42 |
-
- weight_decay : 0
|
| 43 |
-
- lora_dropout : 0
|
| 44 |
- lora_target_modules :['q_proj', 'k_proj', 'v_proj', 'o_proj', 'gate_proj', 'up_proj', 'down_proj']
|
| 45 |
- grad_accum : 8
|
| 46 |
|
|
|
|
| 38 |
- Epochs: 3
|
| 39 |
- Learning rate: 8e-06
|
| 40 |
- LoRA: r=64, alpha=128
|
| 41 |
+
- warmup_ratio : 0.03
|
| 42 |
+
- weight_decay : 0.0
|
| 43 |
+
- lora_dropout : 0.05
|
| 44 |
- lora_target_modules :['q_proj', 'k_proj', 'v_proj', 'o_proj', 'gate_proj', 'up_proj', 'down_proj']
|
| 45 |
- grad_accum : 8
|
| 46 |
|