Upload LoRA adapter (README written by author)
Browse files- README.md +6 -6
- adapter_config.json +6 -6
- adapter_model.safetensors +1 -1
README.md
CHANGED
|
@@ -1,7 +1,7 @@
|
|
| 1 |
---
|
| 2 |
base_model: Qwen/Qwen3-4B-Instruct-2507
|
| 3 |
datasets:
|
| 4 |
-
- daichira/structured-
|
| 5 |
language:
|
| 6 |
- en
|
| 7 |
license: apache-2.0
|
|
@@ -13,7 +13,7 @@ tags:
|
|
| 13 |
- structured-output
|
| 14 |
---
|
| 15 |
|
| 16 |
-
My-
|
| 17 |
|
| 18 |
This repository provides a **LoRA adapter** fine-tuned from
|
| 19 |
**Qwen/Qwen3-4B-Instruct-2507** using **QLoRA (4-bit, Unsloth)**.
|
|
@@ -34,9 +34,9 @@ while intermediate reasoning (Chain-of-Thought) is masked.
|
|
| 34 |
- Base model: Qwen/Qwen3-4B-Instruct-2507
|
| 35 |
- Method: QLoRA (4-bit)
|
| 36 |
- Max sequence length: 1024
|
| 37 |
-
- Epochs:
|
| 38 |
-
- Learning rate:
|
| 39 |
-
- LoRA: r=64, alpha=
|
| 40 |
|
| 41 |
## Usage
|
| 42 |
|
|
@@ -59,7 +59,7 @@ model = PeftModel.from_pretrained(model, adapter)
|
|
| 59 |
|
| 60 |
## Sources & Terms (IMPORTANT)
|
| 61 |
|
| 62 |
-
Training data: daichira/structured-
|
| 63 |
|
| 64 |
Dataset License: MIT License. This dataset is used and distributed under the terms of the MIT License.
|
| 65 |
Compliance: Users must comply with the MIT license (including copyright notice) and the base model's original terms of use.
|
|
|
|
| 1 |
---
|
| 2 |
base_model: Qwen/Qwen3-4B-Instruct-2507
|
| 3 |
datasets:
|
| 4 |
+
- daichira/structured-3k-mix-sft
|
| 5 |
language:
|
| 6 |
- en
|
| 7 |
license: apache-2.0
|
|
|
|
| 13 |
- structured-output
|
| 14 |
---
|
| 15 |
|
| 16 |
+
My-qwen3-4b-structured-output-lora
|
| 17 |
|
| 18 |
This repository provides a **LoRA adapter** fine-tuned from
|
| 19 |
**Qwen/Qwen3-4B-Instruct-2507** using **QLoRA (4-bit, Unsloth)**.
|
|
|
|
| 34 |
- Base model: Qwen/Qwen3-4B-Instruct-2507
|
| 35 |
- Method: QLoRA (4-bit)
|
| 36 |
- Max sequence length: 1024
|
| 37 |
+
- Epochs: 2
|
| 38 |
+
- Learning rate: 5e-05
|
| 39 |
+
- LoRA: r=64, alpha=64
|
| 40 |
|
| 41 |
## Usage
|
| 42 |
|
|
|
|
| 59 |
|
| 60 |
## Sources & Terms (IMPORTANT)
|
| 61 |
|
| 62 |
+
Training data: daichira/structured-3k-mix-sft
|
| 63 |
|
| 64 |
Dataset License: MIT License. This dataset is used and distributed under the terms of the MIT License.
|
| 65 |
Compliance: Users must comply with the MIT license (including copyright notice) and the base model's original terms of use.
|
adapter_config.json
CHANGED
|
@@ -20,7 +20,7 @@
|
|
| 20 |
"layers_pattern": null,
|
| 21 |
"layers_to_transform": null,
|
| 22 |
"loftq_config": {},
|
| 23 |
-
"lora_alpha":
|
| 24 |
"lora_bias": false,
|
| 25 |
"lora_dropout": 0.05,
|
| 26 |
"megatron_config": null,
|
|
@@ -33,13 +33,13 @@
|
|
| 33 |
"rank_pattern": {},
|
| 34 |
"revision": null,
|
| 35 |
"target_modules": [
|
| 36 |
-
"q_proj",
|
| 37 |
-
"k_proj",
|
| 38 |
-
"gate_proj",
|
| 39 |
"up_proj",
|
| 40 |
-
"
|
| 41 |
"v_proj",
|
| 42 |
-
"
|
|
|
|
|
|
|
|
|
|
| 43 |
],
|
| 44 |
"target_parameters": null,
|
| 45 |
"task_type": "CAUSAL_LM",
|
|
|
|
| 20 |
"layers_pattern": null,
|
| 21 |
"layers_to_transform": null,
|
| 22 |
"loftq_config": {},
|
| 23 |
+
"lora_alpha": 64,
|
| 24 |
"lora_bias": false,
|
| 25 |
"lora_dropout": 0.05,
|
| 26 |
"megatron_config": null,
|
|
|
|
| 33 |
"rank_pattern": {},
|
| 34 |
"revision": null,
|
| 35 |
"target_modules": [
|
|
|
|
|
|
|
|
|
|
| 36 |
"up_proj",
|
| 37 |
+
"gate_proj",
|
| 38 |
"v_proj",
|
| 39 |
+
"k_proj",
|
| 40 |
+
"o_proj",
|
| 41 |
+
"down_proj",
|
| 42 |
+
"q_proj"
|
| 43 |
],
|
| 44 |
"target_parameters": null,
|
| 45 |
"task_type": "CAUSAL_LM",
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 528550256
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ed8fa164a7502186a603641e2e7a886f568aef33c1b11bd176c0bf9e0939cce5
|
| 3 |
size 528550256
|