Merge pull request #7366 from oobabooga/dev

Merge dev branch

css/html_instruct_style.css

@@ -19,12 +19,14 @@
     color: #d1d5db !important;
 }
 
-.chat .message-body :is(th, td) {
+.chat .message-body :is(th, td),
+.prose hr {
     border-color: #40404096 !important;
 }
 
-.dark .chat .message-body :is(th, td) {
+.dark .chat .message-body :is(th, td),
-    border-color: #ffffff75 !important;
+.dark .prose hr {
+    border-color: rgb(255 255 255 / 30%) !important;
 }
 
 .chat .message-body :is(p, ul, ol) {

css/main.css

@@ -1797,3 +1797,20 @@ button#swap-height-width {
     top: 0;
     left: calc(100% - 174px);
 }
+
+table {
+    border-collapse: collapse;
+}
+
+table, tr, td, th, thead {
+    border: 0;
+}
+
+td + td,
+th + th { border-left: 1px solid; }
+
+tr + tr td,
+tr + tr th { border-top: 1px solid; }
+
+thead + tbody tr:first-child td,
+thead + tbody tr:first-child th { border-top: 1px solid; }

modules/models_settings.py

@@ -83,7 +83,11 @@ def get_model_metadata(model):
 
         if 'tokenizer.chat_template' in metadata:
             template = metadata['tokenizer.chat_template']
-            eos_token = metadata['tokenizer.ggml.tokens'][metadata['tokenizer.ggml.eos_token_id']]
+            if 'tokenizer.ggml.eos_token_id' in metadata:
+                eos_token = metadata['tokenizer.ggml.tokens'][metadata['tokenizer.ggml.eos_token_id']]
+            else:
+                eos_token = ""
+
             if 'tokenizer.ggml.bos_token_id' in metadata:
                 bos_token = metadata['tokenizer.ggml.tokens'][metadata['tokenizer.ggml.bos_token_id']]
             else:

modules/shared.py

@@ -112,7 +112,7 @@ group.add_argument('--no-cache', action='store_true', help='Set use_cache to Fal
 group.add_argument('--trust-remote-code', action='store_true', help='Set trust_remote_code=True while loading the model. Necessary for some models.')
 group.add_argument('--force-safetensors', action='store_true', help='Set use_safetensors=True while loading the model. This prevents arbitrary code execution.')
 group.add_argument('--no_use_fast', action='store_true', help='Set use_fast=False while loading the tokenizer (it\'s True by default). Use this if you have any problems related to use_fast.')
-group.add_argument('--attn-implementation', type=str, default='flash_attention_2', metavar="IMPLEMENTATION", help='Attention implementation. Valid options: flash_attention_2, sdpa, eager.')
+group.add_argument('--attn-implementation', type=str, default='sdpa', metavar="IMPLEMENTATION", help='Attention implementation. Valid options: sdpa, eager, flash_attention_2.')
 
 # bitsandbytes 4-bit
 group = parser.add_argument_group('bitsandbytes 4-bit')

modules/ui_model_menu.py

@@ -44,7 +44,7 @@ def create_ui():
                             shared.gradio['gpu_layers'] = gr.Slider(label="gpu-layers", minimum=0, maximum=get_initial_gpu_layers_max(), step=1, value=shared.args.gpu_layers, info='Must be greater than 0 for the GPU to be used. ⚠️ Lower this value if you can\'t load the model.')
                             shared.gradio['ctx_size'] = gr.Slider(label='ctx-size', minimum=256, maximum=131072, step=256, value=shared.args.ctx_size, info='Context length. Common values: 4096, 8192, 16384, 32768, 65536, 131072.')
                             shared.gradio['gpu_split'] = gr.Textbox(label='gpu-split', info='Comma-separated list of VRAM (in GB) to use per GPU. Example: 20,7,7')
-                            shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['flash_attention_2', 'sdpa', 'eager'], value=shared.args.attn_implementation, info='Attention implementation.')
+                            shared.gradio['attn_implementation'] = gr.Dropdown(label="attn-implementation", choices=['sdpa', 'eager', 'flash_attention_2'], value=shared.args.attn_implementation, info='Attention implementation.')
                             shared.gradio['cache_type'] = gr.Dropdown(label="cache-type", choices=['fp16', 'q8_0', 'q4_0', 'fp8', 'q8', 'q7', 'q6', 'q5', 'q4', 'q3', 'q2'], value=shared.args.cache_type, allow_custom_value=True, info='Valid options: llama.cpp - fp16, q8_0, q4_0; ExLlamaV2 - fp16, fp8, q8, q6, q4; ExLlamaV3 - fp16, q2 to q8. For ExLlamaV3, you can type custom combinations for separate k/v bits (e.g. q4_q8).')
                             shared.gradio['tp_backend'] = gr.Dropdown(label="tp-backend", choices=['native', 'nccl'], value=shared.args.tp_backend, info='The backend for tensor parallelism.')
 

requirements/full/requirements.txt

@@ -1,11 +1,12 @@
 accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
-bitsandbytes==0.48.*
+bitsandbytes==0.49.*
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
-flash-linear-attention==0.4.0
+flash-linear-attention==0.4.*
 html2text==2025.4.15
 huggingface-hub==0.36.0
 jinja2==3.1.6
@@ -21,13 +22,13 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
@@ -35,19 +36,16 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

requirements/full/requirements_amd.txt

@@ -2,6 +2,7 @@ accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
@@ -19,13 +20,13 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
@@ -33,16 +34,13 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"

requirements/full/requirements_amd_noavx2.txt

@@ -2,6 +2,7 @@ accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
@@ -19,13 +20,13 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
@@ -33,16 +34,13 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+rocm6.2.4.torch2.6.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"

requirements/full/requirements_apple_intel.txt

@@ -2,6 +2,7 @@ accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
@@ -19,13 +20,13 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
@@ -33,14 +34,11 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"

requirements/full/requirements_apple_silicon.txt

@@ -2,6 +2,7 @@ accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
@@ -19,13 +20,13 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
@@ -33,14 +34,11 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"

requirements/full/requirements_cpu_only.txt

@@ -2,6 +2,7 @@ accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
@@ -19,13 +20,13 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
@@ -33,14 +34,11 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"

requirements/full/requirements_cpu_only_noavx2.txt

@@ -2,6 +2,7 @@ accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
@@ -19,13 +20,13 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
@@ -33,14 +34,11 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"

requirements/full/requirements_noavx2.txt

@@ -1,11 +1,12 @@
 accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
-bitsandbytes==0.48.*
+bitsandbytes==0.49.*
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
-flash-linear-attention==0.4.0
+flash-linear-attention==0.4.*
 html2text==2025.4.15
 huggingface-hub==0.36.0
 jinja2==3.1.6
@@ -21,13 +22,13 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
@@ -35,19 +36,16 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/turboderp-org/exllamav3/releases/download/v0.0.17/exllamav3-0.0.17+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/turboderp-org/exllamav3/releases/download/v0.0.18/exllamav3-0.0.18+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2+cu128.torch2.7.0-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/turboderp-org/exllamav2/releases/download/v0.3.2/exllamav2-0.3.2-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"

requirements/full/requirements_nowheels.txt

@@ -2,6 +2,7 @@ accelerate==1.8.*
 audioop-lts<1.0; python_version >= "3.13"
 colorama
 datasets
+diffusers==0.36.*
 einops
 fastapi==0.112.4
 html2text==2025.4.15
@@ -19,13 +20,13 @@ python-docx==1.1.2
 pyyaml
 requests
 rich
-safetensors==0.6.*
+safetensors==0.7.*
 scipy
 sentencepiece
 tensorboard
-torchao==0.14.*
+torchao==0.15.*
 transformers==4.57.*
-triton-windows==3.5.1.post21; platform_system == "Windows"
+triton-windows==3.5.1.post22; platform_system == "Windows"
 tqdm
 wandb
 
@@ -33,9 +34,6 @@ wandb
 gradio==4.37.*
 https://github.com/oobabooga/gradio/releases/download/custom-build/gradio_client-1.0.2+custom.1-py3-none-any.whl
 
-# Diffusers
-diffusers @ git+https://github.com/huggingface/diffusers.git@edf36f5128abf3e6ecf92b5145115514363c58e6
-
 # API
 flask_cloudflared==0.0.14
 sse-starlette==1.6.5

requirements/portable/requirements.txt

@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

requirements/portable/requirements_amd.txt

@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+rocm6.4.4-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

requirements/portable/requirements_amd_noavx2.txt

@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # AMD wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+rocm6.4.4avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

requirements/portable/requirements_apple_intel.txt

@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_15_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_14_0_x86_64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"

requirements/portable/requirements_apple_silicon.txt

@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Mac wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_15_0_arm64.whl; platform_system == "Darwin" and platform_release >= "24.0.0"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0-py3-none-macosx_14_0_arm64.whl; platform_system == "Darwin" and platform_release >= "23.0.0" and platform_release < "24.0.0"

requirements/portable/requirements_cpu_only.txt

@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx2-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx2-py3-none-win_amd64.whl; platform_system == "Windows"

requirements/portable/requirements_cpu_only_noavx2.txt

@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # llama.cpp (CPU only, no AVX2)
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cpuavx-py3-none-win_amd64.whl; platform_system == "Windows"

requirements/portable/requirements_noavx2.txt

@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+cu124avx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

requirements/portable/requirements_vulkan.txt

@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # Vulkan wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkan-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkan-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"

requirements/portable/requirements_vulkan_noavx2.txt

@@ -23,5 +23,5 @@ sse-starlette==1.6.5
 tiktoken
 
 # CUDA wheels
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-win_amd64.whl; platform_system == "Windows"
-https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.66.0/llama_cpp_binaries-0.66.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"
+https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.74.0/llama_cpp_binaries-0.74.0+vulkanavx-py3-none-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64"