Merge branch 'master' into fix/stream_tokens_usage

Makefile

@@ -8,7 +8,7 @@ DETECT_LIBS?=true
 # llama.cpp versions
 GOLLAMA_REPO?=https://github.com/go-skynet/go-llama.cpp
 GOLLAMA_VERSION?=2b57a8ae43e4699d3dc5d1496a1ccd42922993be
-CPPLLAMA_VERSION?=30ec39832165627dd6ed98938df63adfc6e6a21a
+CPPLLAMA_VERSION?=3ad5451f3b75809e3033e4e577b9f60bcaf6676a
 
 # whisper.cpp version
 WHISPER_REPO?=https://github.com/ggerganov/whisper.cpp

gallery/index.yaml

@@ -1614,6 +1614,23 @@
     - filename: Q2.5-MS-Mistoria-72b-v2-Q4_K_M.gguf
       sha256: 33df8aac5a790d1c286fe0fc4f9d340311f282eca19b78db6f7abb845923425c
       uri: huggingface://bartowski/Q2.5-MS-Mistoria-72b-v2-GGUF/Q2.5-MS-Mistoria-72b-v2-Q4_K_M.gguf
+- !!merge <<: *qwen25
+  name: "eva-qwen2.5-72b-v0.2"
+  urls:
+    - https://huggingface.co/EVA-UNIT-01/EVA-Qwen2.5-72B-v0.2
+    - https://huggingface.co/bartowski/EVA-Qwen2.5-72B-v0.2-GGUF
+  description: |
+    A RP/storywriting specialist model, full-parameter finetune of Qwen2.5-72B on mixture of synthetic and natural data.
+    It uses Celeste 70B 0.1 data mixture, greatly expanding it to improve versatility, creativity and "flavor" of the resulting model.
+
+    Version notes for 0.2: Optimized training hyperparameters and increased sequence length. Better instruction following deeper into context and less repetition.
+  overrides:
+    parameters:
+      model: EVA-Qwen2.5-72B-v0.2-Q4_K_M.gguf
+  files:
+    - filename: EVA-Qwen2.5-72B-v0.2-Q4_K_M.gguf
+      sha256: 03ea0ecac3ee24a332ca43cf925b669c58714b9754be0f4bc232bd996681ef4b
+      uri: huggingface://bartowski/EVA-Qwen2.5-72B-v0.2-GGUF/EVA-Qwen2.5-72B-v0.2-Q4_K_M.gguf
 - &archfunct
   license: apache-2.0
   tags:
@@ -3176,6 +3193,36 @@
     - filename: Tulu-3.1-8B-SuperNova.i1-Q4_K_M.gguf
       sha256: c6cc2e1a4c3d2338973ca0050af1cf4462b3f62838f62b4c8a204f2a74eeb01f
       uri: huggingface://mradermacher/Tulu-3.1-8B-SuperNova-i1-GGUF/Tulu-3.1-8B-SuperNova.i1-Q4_K_M.gguf
+- !!merge <<: *llama31
+  name: "llama-3.1-tulu-3-70b-dpo"
+  icon: "https://huggingface.co/datasets/allenai/blog-images/resolve/main/tulu3/Tulu3-logo.png"
+  urls:
+    - https://huggingface.co/allenai/Llama-3.1-Tulu-3-70B-DPO
+    - https://huggingface.co/bartowski/Llama-3.1-Tulu-3-70B-DPO-GGUF
+  description: |
+    Tülu3 is a leading instruction following model family, offering fully open-source data, code, and recipes designed to serve as a comprehensive guide for modern post-training techniques. Tülu3 is designed for state-of-the-art performance on a diversity of tasks in addition to chat, such as MATH, GSM8K, and IFEval.
+  overrides:
+    parameters:
+      model: Llama-3.1-Tulu-3-70B-DPO-Q4_K_M.gguf
+  files:
+    - filename: Llama-3.1-Tulu-3-70B-DPO-Q4_K_M.gguf
+      sha256: e2d9c59736274f9dd94f30ef3edcee68fec1d6649eb01d6bad7e3e8a6024f77d
+      uri: huggingface://bartowski/Llama-3.1-Tulu-3-70B-DPO-GGUF/Llama-3.1-Tulu-3-70B-DPO-Q4_K_M.gguf
+- !!merge <<: *llama31
+  name: "llama-3.1-tulu-3-8b-sft"
+  icon: "https://huggingface.co/datasets/allenai/blog-images/resolve/main/tulu3/Tulu3-logo.png"
+  urls:
+    - https://huggingface.co/allenai/Llama-3.1-Tulu-3-8B-SFT
+    - https://huggingface.co/bartowski/Llama-3.1-Tulu-3-8B-SFT-GGUF
+  description: |
+    Tülu3 is a leading instruction following model family, offering fully open-source data, code, and recipes designed to serve as a comprehensive guide for modern post-training techniques. Tülu3 is designed for state-of-the-art performance on a diversity of tasks in addition to chat, such as MATH, GSM8K, and IFEval.
+  overrides:
+    parameters:
+      model: Llama-3.1-Tulu-3-8B-SFT-Q4_K_M.gguf
+  files:
+    - filename: Llama-3.1-Tulu-3-8B-SFT-Q4_K_M.gguf
+      sha256: 3fad2c96aa9b9de19c2cda0f88a381c47ac768ca03a95059d9f6c439791f8592
+      uri: huggingface://bartowski/Llama-3.1-Tulu-3-8B-SFT-GGUF/Llama-3.1-Tulu-3-8B-SFT-Q4_K_M.gguf
 - &deepseek
   ## Deepseek
   url: "github:mudler/LocalAI/gallery/deepseek.yaml@master"