bump 3.5 (#2565)

Author: vince62s

CHANGELOG.md

@@ -3,6 +3,17 @@
 
 ## [Unreleased]
 
+## [3.5.0](https://github.com/OpenNMT/OpenNMT-py/tree/3.5.0) (2024-02-22)
+
+* Further improvements and fixes
+* Suport for AWQ models
+* Add n_best for topp/topk generation
+* Support MoE (MIxtral) inference
+* Extend HF models converter
+* use flash_attn_with_kvcache for faster inference
+* Add wikitext2 PPL computation
+* Support for Phi-2 models
+
 ## [3.4.3](https://github.com/OpenNMT/OpenNMT-py/tree/3.4.3) (2023-11-2)
 
 * Further improvements to beam search and decoding

README.md

@@ -25,7 +25,7 @@ Otherwise you can just have a look at the [Quickstart](https://opennmt.net/OpenN
 ----
 ## New:
 
-* You will need Pytorch v2 preferably v2.1 which fixes some `scaled_dot_product_attention` issues
+* You will need Pytorch v2 preferably v2.2 which fixes some `scaled_dot_product_attention` issues
 * LLM support with converters for: Llama (+ Mistral), OpenLlama, Redpajama, MPT-7B, Falcon.
 * Support for 8bit and 4bit quantization along with LoRA adapters, with or without checkpointing.
 * You can finetune 7B and 13B models on a single RTX 24GB with 4-bit quantization.

onmt/__init__.py

@@ -21,4 +21,4 @@
     onmt.modules,
 ]
 
-__version__ = "3.4.3"
+__version__ = "3.5.0"

setup.py

@@ -11,7 +11,7 @@
     description="A python implementation of OpenNMT",
     long_description=long_description,
     long_description_content_type="text/markdown",
-    version="3.4.3",
+    version="3.5.0",
     packages=find_packages(),
     project_urls={
         "Documentation": "http://opennmt.net/OpenNMT-py/",
@@ -21,9 +21,9 @@
     },
     python_requires=">=3.8",
     install_requires=[
-        "torch>=2.0.1,<2.2",
+        "torch>=2.0.1,<2.3",
         "configargparse",
-        "ctranslate2>=3.17,<4",
+        "ctranslate2>=3.24,<5",
         "tensorboard>=2.3",
         "flask",
         "waitress",