Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
# Model Card: Qwen3-Embedding-0.6B Fine-tuned with LoRA
|
2 |
|
3 |
## Model Details
|
@@ -10,11 +36,15 @@
|
|
10 |
* **Contact:** [Your Email/Contact Information Here]
|
11 |
* **Date:** July 13, 2025
|
12 |
|
|
|
|
|
13 |
## Model Description
|
14 |
|
15 |
-
This model is a fine-tuned version of the Qwen3-Embedding-0.6B model, adapted using the LoRA method. The goal of this fine-tuning was to enhance its performance on specific downstream tasks (e.g., semantic search, clustering, recommendation systems) by aligning its embeddings more closely with the characteristics of a particular dataset.
|
|
|
|
|
16 |
|
17 |
-
|
18 |
|
19 |
## Intended Use
|
20 |
|
@@ -26,6 +56,8 @@ This model is intended for:
|
|
26 |
* Information retrieval and recommendation systems.
|
27 |
* As a component in larger NLP pipelines where robust text representations are required.
|
28 |
|
|
|
|
|
29 |
## Limitations and Biases
|
30 |
|
31 |
* **Domain Specificity:** While fine-tuned, the model's performance may degrade on data significantly different from its training distribution.
|
@@ -33,6 +65,8 @@ This model is intended for:
|
|
33 |
* **Computational Resources:** While LoRA reduces resource demands for fine-tuning, inference still requires appropriate computational resources.
|
34 |
* **Language:** Primarily designed for [Specify Language(s) if known, e.g., English] text. Performance on other languages may vary.
|
35 |
|
|
|
|
|
36 |
## Training Details
|
37 |
|
38 |
* **Fine-tuning Method:** LoRA (Low-Rank Adaptation)
|
@@ -47,6 +81,8 @@ This model is intended for:
|
|
47 |
* **Optimization Strategy:** [e.g., AdamW, learning rate schedule]
|
48 |
* **Software Frameworks:** [e.g., PyTorch, Hugging Face Transformers, PEFT library]
|
49 |
|
|
|
|
|
50 |
## Performance Metrics
|
51 |
|
52 |
*(Note: Provide actual metrics from your evaluation. Examples below are placeholders.)*
|
@@ -57,6 +93,8 @@ This model is intended for:
|
|
57 |
* **Metric 3 (e.g., Cosine Similarity Distribution):** [Description or relevant statistics]
|
58 |
* **Comparison to Base Model (if available):** [e.g., "This fine-tuned model showed a 15% improvement in Average Precision @ 10 compared to the base Qwen3-Embedding-0.6B model on our internal benchmark."]
|
59 |
|
|
|
|
|
60 |
## Usage
|
61 |
|
62 |
You can load and use this model with the Hugging Face `transformers` and `peft` libraries.
|
@@ -109,17 +147,3 @@ If you use this fine-tuned model in your research or application, please conside
|
|
109 |
year={2025},
|
110 |
note={Available at [Link to your model if uploaded]}
|
111 |
}
|
112 |
-
|
113 |
-
---
|
114 |
-
|
115 |
-
## License
|
116 |
-
|
117 |
-
This fine-tuned model inherits the license of the original **Qwen/Qwen3-Embedding-0.6B** model. Please refer to the [original model's license]([Link to original model's license, e.g., Hugging Face model page]) for details.
|
118 |
-
|
119 |
-
---
|
120 |
-
|
121 |
-
## Acknowledgements
|
122 |
-
|
123 |
-
* The developers of **Qwen/Qwen3-Embedding-0.6B** for providing the base model.
|
124 |
-
* The developers of the **PEFT** library for enabling efficient LoRA fine-tuning.
|
125 |
-
* [Any other relevant acknowledgements, e.g., dataset creators, funding bodies]
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0 # Or your model's specific license, e.g., mit, gpl-3.0, custom
|
3 |
+
tags:
|
4 |
+
- text-embedding
|
5 |
+
- qwen
|
6 |
+
- lora
|
7 |
+
- fine-tuning
|
8 |
+
- representation-learning
|
9 |
+
language: en # Example, adjust if your model is for other languages
|
10 |
+
model-index:
|
11 |
+
- name: qwen3-embedding-0.6b-lora-fine-tuned
|
12 |
+
results:
|
13 |
+
- task:
|
14 |
+
type: text-embedding
|
15 |
+
name: Text Embedding
|
16 |
+
dataset:
|
17 |
+
name: "Semantic Similar Dataset"
|
18 |
+
type: "Semantic"
|
19 |
+
metrics:
|
20 |
+
- type: average_precision # Use a standard metric identifier if possible
|
21 |
+
value: 0.85 # Your model's score for this metric
|
22 |
+
name: Average Precision @ K
|
23 |
+
- type: recall
|
24 |
+
value: 0.92
|
25 |
+
name: Recall @ K
|
26 |
+
---
|
27 |
# Model Card: Qwen3-Embedding-0.6B Fine-tuned with LoRA
|
28 |
|
29 |
## Model Details
|
|
|
36 |
* **Contact:** [Your Email/Contact Information Here]
|
37 |
* **Date:** July 13, 2025
|
38 |
|
39 |
+
---
|
40 |
+
|
41 |
## Model Description
|
42 |
|
43 |
+
This model is a fine-tuned version of the **Qwen3-Embedding-0.6B** model, adapted using the **LoRA** method. The goal of this fine-tuning was to enhance its performance on specific downstream tasks (e.g., semantic search, clustering, recommendation systems) by aligning its embeddings more closely with the characteristics of a particular dataset.
|
44 |
+
|
45 |
+
**Qwen3-Embedding-0.6B** is an efficient and performant embedding model from the Qwen series, designed to convert text into high-dimensional numerical vectors (embeddings) that capture semantic meaning. **LoRA** fine-tuning allows for efficient adaptation of large pre-trained models with minimal computational cost and storage requirements, making it ideal for targeted performance improvements without full model retraining.
|
46 |
|
47 |
+
---
|
48 |
|
49 |
## Intended Use
|
50 |
|
|
|
56 |
* Information retrieval and recommendation systems.
|
57 |
* As a component in larger NLP pipelines where robust text representations are required.
|
58 |
|
59 |
+
---
|
60 |
+
|
61 |
## Limitations and Biases
|
62 |
|
63 |
* **Domain Specificity:** While fine-tuned, the model's performance may degrade on data significantly different from its training distribution.
|
|
|
65 |
* **Computational Resources:** While LoRA reduces resource demands for fine-tuning, inference still requires appropriate computational resources.
|
66 |
* **Language:** Primarily designed for [Specify Language(s) if known, e.g., English] text. Performance on other languages may vary.
|
67 |
|
68 |
+
---
|
69 |
+
|
70 |
## Training Details
|
71 |
|
72 |
* **Fine-tuning Method:** LoRA (Low-Rank Adaptation)
|
|
|
81 |
* **Optimization Strategy:** [e.g., AdamW, learning rate schedule]
|
82 |
* **Software Frameworks:** [e.g., PyTorch, Hugging Face Transformers, PEFT library]
|
83 |
|
84 |
+
---
|
85 |
+
|
86 |
## Performance Metrics
|
87 |
|
88 |
*(Note: Provide actual metrics from your evaluation. Examples below are placeholders.)*
|
|
|
93 |
* **Metric 3 (e.g., Cosine Similarity Distribution):** [Description or relevant statistics]
|
94 |
* **Comparison to Base Model (if available):** [e.g., "This fine-tuned model showed a 15% improvement in Average Precision @ 10 compared to the base Qwen3-Embedding-0.6B model on our internal benchmark."]
|
95 |
|
96 |
+
---
|
97 |
+
|
98 |
## Usage
|
99 |
|
100 |
You can load and use this model with the Hugging Face `transformers` and `peft` libraries.
|
|
|
147 |
year={2025},
|
148 |
note={Available at [Link to your model if uploaded]}
|
149 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|