Upload README.md
Browse files
README.md
CHANGED
|
@@ -117,12 +117,12 @@ Most GPTQ files are made with AutoGPTQ. Mistral models are currently made with T
|
|
| 117 |
|
| 118 |
| Branch | Bits | GS | Act Order | Damp % | GPTQ Dataset | Seq Len | Size | ExLlama | Desc |
|
| 119 |
| ------ | ---- | -- | --------- | ------ | ------------ | ------- | ---- | ------- | ---- |
|
| 120 |
-
| [main](https://huggingface.co/TheBloke/Synatra-7B-v0.3-base-GPTQ/tree/main) | 4 | 128 | Yes | 0.1 | [
|
| 121 |
-
| [gptq-4bit-32g-actorder_True](https://huggingface.co/TheBloke/Synatra-7B-v0.3-base-GPTQ/tree/gptq-4bit-32g-actorder_True) | 4 | 32 | Yes | 0.1 | [
|
| 122 |
-
| [gptq-8bit--1g-actorder_True](https://huggingface.co/TheBloke/Synatra-7B-v0.3-base-GPTQ/tree/gptq-8bit--1g-actorder_True) | 8 | None | Yes | 0.1 | [
|
| 123 |
-
| [gptq-8bit-128g-actorder_True](https://huggingface.co/TheBloke/Synatra-7B-v0.3-base-GPTQ/tree/gptq-8bit-128g-actorder_True) | 8 | 128 | Yes | 0.1 | [
|
| 124 |
-
| [gptq-8bit-32g-actorder_True](https://huggingface.co/TheBloke/Synatra-7B-v0.3-base-GPTQ/tree/gptq-8bit-32g-actorder_True) | 8 | 32 | Yes | 0.1 | [
|
| 125 |
-
| [gptq-4bit-64g-actorder_True](https://huggingface.co/TheBloke/Synatra-7B-v0.3-base-GPTQ/tree/gptq-4bit-64g-actorder_True) | 4 | 64 | Yes | 0.1 | [
|
| 126 |
|
| 127 |
<!-- README_GPTQ.md-provided-files end -->
|
| 128 |
|
|
|
|
| 117 |
|
| 118 |
| Branch | Bits | GS | Act Order | Damp % | GPTQ Dataset | Seq Len | Size | ExLlama | Desc |
|
| 119 |
| ------ | ---- | -- | --------- | ------ | ------------ | ------- | ---- | ------- | ---- |
|
| 120 |
+
| [main](https://huggingface.co/TheBloke/Synatra-7B-v0.3-base-GPTQ/tree/main) | 4 | 128 | Yes | 0.1 | [Korean Alpaca](https://huggingface.co/datasets/beomi/KoAlpaca-v1.1a/viewer/) | 4096 | 4.16 GB | Yes | 4-bit, with Act Order and group size 128g. Uses even less VRAM than 64g, but with slightly lower accuracy. |
|
| 121 |
+
| [gptq-4bit-32g-actorder_True](https://huggingface.co/TheBloke/Synatra-7B-v0.3-base-GPTQ/tree/gptq-4bit-32g-actorder_True) | 4 | 32 | Yes | 0.1 | [Korean Alpaca](https://huggingface.co/datasets/beomi/KoAlpaca-v1.1a/viewer/) | 4096 | 4.57 GB | Yes | 4-bit, with Act Order and group size 32g. Gives highest possible inference quality, with maximum VRAM usage. |
|
| 122 |
+
| [gptq-8bit--1g-actorder_True](https://huggingface.co/TheBloke/Synatra-7B-v0.3-base-GPTQ/tree/gptq-8bit--1g-actorder_True) | 8 | None | Yes | 0.1 | [Korean Alpaca](https://huggingface.co/datasets/beomi/KoAlpaca-v1.1a/viewer/) | 4096 | 7.52 GB | No | 8-bit, with Act Order. No group size, to lower VRAM requirements. |
|
| 123 |
+
| [gptq-8bit-128g-actorder_True](https://huggingface.co/TheBloke/Synatra-7B-v0.3-base-GPTQ/tree/gptq-8bit-128g-actorder_True) | 8 | 128 | Yes | 0.1 | [Korean Alpaca](https://huggingface.co/datasets/beomi/KoAlpaca-v1.1a/viewer/) | 4096 | 7.68 GB | No | 8-bit, with group size 128g for higher inference quality and with Act Order for even higher accuracy. |
|
| 124 |
+
| [gptq-8bit-32g-actorder_True](https://huggingface.co/TheBloke/Synatra-7B-v0.3-base-GPTQ/tree/gptq-8bit-32g-actorder_True) | 8 | 32 | Yes | 0.1 | [Korean Alpaca](https://huggingface.co/datasets/beomi/KoAlpaca-v1.1a/viewer/) | 4096 | 8.17 GB | No | 8-bit, with group size 32g and Act Order for maximum inference quality. |
|
| 125 |
+
| [gptq-4bit-64g-actorder_True](https://huggingface.co/TheBloke/Synatra-7B-v0.3-base-GPTQ/tree/gptq-4bit-64g-actorder_True) | 4 | 64 | Yes | 0.1 | [Korean Alpaca](https://huggingface.co/datasets/beomi/KoAlpaca-v1.1a/viewer/) | 4096 | 4.30 GB | Yes | 4-bit, with Act Order and group size 64g. Uses less VRAM than 32g, but with slightly lower accuracy. |
|
| 126 |
|
| 127 |
<!-- README_GPTQ.md-provided-files end -->
|
| 128 |
|