Add costs breakdown
Browse files
README.md
CHANGED
|
@@ -27,10 +27,6 @@ tags:
|
|
| 27 |
- **MultiPL-E (C++)**: 32.92 %
|
| 28 |
- **Average**: 34.38 %
|
| 29 |
|
| 30 |
-
## Communication Budget
|
| 31 |
-
|
| 32 |
-
8766.51 MB Megabytes
|
| 33 |
-
|
| 34 |
## Model Details
|
| 35 |
|
| 36 |
This PEFT adapter has been trained by using [Flower](https://flower.ai/), a friendly federated AI framework.
|
|
@@ -53,6 +49,56 @@ base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen2.5-Coder-0.5B-Instr
|
|
| 53 |
model = PeftModel.from_pretrained(base_model, "ethicalabs/FlowerTune-Qwen2.5-Coder-0.5B-Instruct")
|
| 54 |
```
|
| 55 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 56 |
### Framework versions
|
| 57 |
|
| 58 |
- PEFT 0.14.0
|
|
|
|
| 27 |
- **MultiPL-E (C++)**: 32.92 %
|
| 28 |
- **Average**: 34.38 %
|
| 29 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 30 |
## Model Details
|
| 31 |
|
| 32 |
This PEFT adapter has been trained by using [Flower](https://flower.ai/), a friendly federated AI framework.
|
|
|
|
| 49 |
model = PeftModel.from_pretrained(base_model, "ethicalabs/FlowerTune-Qwen2.5-Coder-0.5B-Instruct")
|
| 50 |
```
|
| 51 |
|
| 52 |
+
## Communication Budget
|
| 53 |
+
|
| 54 |
+
8766.51 MB Megabytes
|
| 55 |
+
|
| 56 |
+
## Virtual Machine Details
|
| 57 |
+
|
| 58 |
+
For this experiment, I utilized [CUDO Compute](https://www.cudocompute.com/?via=flowertune-llm) as the GPU compute provider.
|
| 59 |
+
|
| 60 |
+
| **Component** | **Specification** |
|
| 61 |
+
|---------------|----------------------|
|
| 62 |
+
| **GPU** | 1 × RTX A4000 16 GB |
|
| 63 |
+
| **vCPUs** | 4 |
|
| 64 |
+
| **CPU** | AMD EPYC (Milan) |
|
| 65 |
+
| **Memory** | 16 GB |
|
| 66 |
+
|
| 67 |
+
## Cost Breakdown
|
| 68 |
+
|
| 69 |
+
### Compute Costs
|
| 70 |
+
|
| 71 |
+
| **Component** | **Details** | **Cost/hr** |
|
| 72 |
+
|---------------|---------------|-------------|
|
| 73 |
+
| vCPUs | 4 cores | $0.0088/hr |
|
| 74 |
+
| Memory | 16 GB | $0.056/hr |
|
| 75 |
+
| GPU | 1 × RTX A4000 | $0.25/hr |
|
| 76 |
+
|
| 77 |
+
### Storage Costs
|
| 78 |
+
|
| 79 |
+
| **Component** | **Details** | **Cost/hr** |
|
| 80 |
+
|------------------|-------------|-------------|
|
| 81 |
+
| Boot Disk Size | 70 GB | $0.0077/hr |
|
| 82 |
+
|
| 83 |
+
### Network Costs
|
| 84 |
+
|
| 85 |
+
| **Component** | **Details** | **Cost/hr** |
|
| 86 |
+
|-----------------------|-------------|-------------|
|
| 87 |
+
| Public IPv4 Address | N/A | $0.005/hr |
|
| 88 |
+
|
| 89 |
+
### Total Cost
|
| 90 |
+
|
| 91 |
+
| **Total Cost/hr** |
|
| 92 |
+
|-------------------|
|
| 93 |
+
| **$0.3275/hr** |
|
| 94 |
+
|
| 95 |
+
### Simulation Details
|
| 96 |
+
|
| 97 |
+
| **Parameter** | **Value** |
|
| 98 |
+
|--------------------|------------------------|
|
| 99 |
+
| **Runtime** | 1924.52 seconds (00:32:04) |
|
| 100 |
+
| **Simulation Cost**| **$0.18** |
|
| 101 |
+
|
| 102 |
### Framework versions
|
| 103 |
|
| 104 |
- PEFT 0.14.0
|