Update README.md
Browse files
README.md
CHANGED
@@ -109,7 +109,27 @@ pipe = setup_pipeline(llm_path, use_8bit=True)
|
|
109 |
## Training process
|
110 |
**Training Metrics Image**: Below is a snapshot of the training loss curve visualized.
|
111 |
|
112 |
-
![
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
113 |
|
114 |
### Hardware
|
115 |
|
|
|
109 |
## Training process
|
110 |
**Training Metrics Image**: Below is a snapshot of the training loss curve visualized.
|
111 |
|
112 |
+
![image/png](https://cdn-uploads.huggingface.co/production/uploads/65713d70f56f9538679e5a56/DmZOYY_-NQtNS610HXR8L.png)
|
113 |
+
|
114 |
+
**[MMLU](https://huggingface.co/datasets/cais/mmlu)**:
|
115 |
+
|
116 |
+
| Model | MMLU Score |
|
117 |
+
| --- | --- |
|
118 |
+
| llama3.5-instruct-8b | 69.40 |
|
119 |
+
| ichigo-llama3.1-s-v0.3: phase 3 | 63.79 |
|
120 |
+
| ichigo-llama3.1-s-v0.3: phase 2 | **63.08** |
|
121 |
+
| ichigo-llama3.1-s-base-v0.3 | 42.11 |
|
122 |
+
| llama3.5-instruct-v0.2 | 50.27 |
|
123 |
+
|
124 |
+
**[AudioBench](https://arxiv.org/abs/2406.16020) Eval**:
|
125 |
+
|
126 |
+
| Model Bench | [Open-hermes Instruction Audio](https://huggingface.co/datasets/AudioLLMs/openhermes_instruction_test) (GPT-4-O judge 0:5) | [Alpaca Instruction Audio](https://huggingface.co/datasets/AudioLLMs/alpaca_audio_test) (GPT-4-O judge 0:5) |
|
127 |
+
| --- | --- | --- |
|
128 |
+
| [Llama3.1-s-v2](https://huggingface.co/homebrewltd/llama3-s-instruct-v0.2) | 3.45 | 3.53 |
|
129 |
+
| [Ichigo-llama3.1-s v0.3-phase2 -cp7000](https://huggingface.co/homebrewltd/Ichigo-llama3.1-s-instruct-v0.3-phase-2) | **3.42** | **3.62** |
|
130 |
+
| [Ichigo-llama3.1-s v0.3-phase2-cplast](https://huggingface.co/jan-hq/llama3-s-instruct-v0.3-checkpoint-last) | 3.31 | 3.6 |
|
131 |
+
| [Ichigo-llama3.1-s v0.3-phase3](https://huggingface.co/homebrewltd/Ichigo-llama3.1-s-instruct-v0.3-phase-3) | 3.64 | 3.68 |
|
132 |
+
| [Qwen2-audio-7B](https://huggingface.co/Qwen/Qwen2-Audio-7B) | 2.63 | 2.24 |
|
133 |
|
134 |
### Hardware
|
135 |
|