Merge branch 'master' into gw-list-model-filter-upgrade

mudler · Sep 27, 2024 · 1b50daa · 1b50daa
2 parents f9dd9d2 + 2a8cbad
commit 1b50daa
Showing 1 changed file with 54 additions and 17 deletions.
diff --git a/gallery/index.yaml b/gallery/index.yaml
@@ -59,8 +59,8 @@
     - filename: llama-3.2-1b-instruct-q8_0.gguf
       sha256: ba345c83bf5cc679c653b853c46517eea5a34f03ed2205449db77184d9ae62a9
       uri: huggingface://hugging-quants/Llama-3.2-1B-Instruct-Q8_0-GGUF/llama-3.2-1b-instruct-q8_0.gguf
-## Qwen2.5
 - &qwen25
+  ## Qwen2.5
   name: "qwen2.5-14b-instruct"
   url: "github:mudler/LocalAI/gallery/chatml.yaml@master"
   license: apache-2.0
@@ -89,11 +89,11 @@
     - https://huggingface.co/bartowski/Qwen2.5-Math-7B-Instruct-GGUF
     - https://huggingface.co/Qwen/Qwen2.5-Math-7B-Instruct
   description: |
-      In August 2024, we released the first series of mathematical LLMs - Qwen2-Math - of our Qwen family. A month later, we have upgraded it and open-sourced Qwen2.5-Math series, including base models Qwen2.5-Math-1.5B/7B/72B, instruction-tuned models Qwen2.5-Math-1.5B/7B/72B-Instruct, and mathematical reward model Qwen2.5-Math-RM-72B.
+    In August 2024, we released the first series of mathematical LLMs - Qwen2-Math - of our Qwen family. A month later, we have upgraded it and open-sourced Qwen2.5-Math series, including base models Qwen2.5-Math-1.5B/7B/72B, instruction-tuned models Qwen2.5-Math-1.5B/7B/72B-Instruct, and mathematical reward model Qwen2.5-Math-RM-72B.
 
-      Unlike Qwen2-Math series which only supports using Chain-of-Thught (CoT) to solve English math problems, Qwen2.5-Math series is expanded to support using both CoT and Tool-integrated Reasoning (TIR) to solve math problems in both Chinese and English. The Qwen2.5-Math series models have achieved significant performance improvements compared to the Qwen2-Math series models on the Chinese and English mathematics benchmarks with CoT.
+    Unlike Qwen2-Math series which only supports using Chain-of-Thught (CoT) to solve English math problems, Qwen2.5-Math series is expanded to support using both CoT and Tool-integrated Reasoning (TIR) to solve math problems in both Chinese and English. The Qwen2.5-Math series models have achieved significant performance improvements compared to the Qwen2-Math series models on the Chinese and English mathematics benchmarks with CoT.
 
-      The base models of Qwen2-Math are initialized with Qwen2-1.5B/7B/72B, and then pretrained on a meticulously designed Mathematics-specific Corpus. This corpus contains large-scale high-quality mathematical web texts, books, codes, exam questions, and mathematical pre-training data synthesized by Qwen2.
+    The base models of Qwen2-Math are initialized with Qwen2-1.5B/7B/72B, and then pretrained on a meticulously designed Mathematics-specific Corpus. This corpus contains large-scale high-quality mathematical web texts, books, codes, exam questions, and mathematical pre-training data synthesized by Qwen2.
   overrides:
     parameters:
       model: Qwen2.5-Math-7B-Instruct-Q4_K_M.gguf
@@ -195,8 +195,8 @@
       model: Qwen2.5-32B.Q4_K_M.gguf
   files:
     - filename: Qwen2.5-32B.Q4_K_M.gguf
-      sha256: 02703e27c8b964db445444581a6937ad7538f0c32a100b26b49fa0e8ff527155
       uri: huggingface://mradermacher/Qwen2.5-32B-GGUF/Qwen2.5-32B.Q4_K_M.gguf
+      sha256: fa42a4067e3630929202b6bb1ef5cebc43c1898494aedfd567b7d53c7a9d84a6
 - !!merge <<: *qwen25
   name: "qwen2.5-32b-instruct"
   urls:
@@ -221,8 +221,24 @@
     - filename: Qwen2.5-72B-Instruct-Q4_K_M.gguf
       sha256: e4c8fad16946be8cf0bbf67eb8f4e18fc7415a5a6d2854b4cda453edb4082545
       uri: huggingface://bartowski/Qwen2.5-72B-Instruct-GGUF/Qwen2.5-72B-Instruct-Q4_K_M.gguf
-## SmolLM
+- !!merge <<: *qwen25
+  name: "bigqwen2.5-52b-instruct"
+  icon: https://cdn-uploads.huggingface.co/production/uploads/61b8e2ba285851687028d395/98GiKtmH1AtHHbIbOUH4Y.jpeg
+  urls:
+    - https://huggingface.co/mlabonne/BigQwen2.5-52B-Instruct
+    - https://huggingface.co/bartowski/BigQwen2.5-52B-Instruct-GGUF
+  description: |
+    BigQwen2.5-52B-Instruct is a Qwen/Qwen2-32B-Instruct self-merge made with MergeKit.
+    It applies the mlabonne/Meta-Llama-3-120B-Instruct recipe.
+  overrides:
+    parameters:
+      model: BigQwen2.5-52B-Instruct-Q4_K_M.gguf
+  files:
+    - filename: BigQwen2.5-52B-Instruct-Q4_K_M.gguf
+      sha256: 9c939f08e366b51b07096eb2ecb5cc2a82894ac7baf639e446237ad39889c896
+      uri: huggingface://bartowski/BigQwen2.5-52B-Instruct-GGUF/BigQwen2.5-52B-Instruct-Q4_K_M.gguf
 - &smollm
+  ## SmolLM
   url: "github:mudler/LocalAI/gallery/chatml.yaml@master"
   name: "smollm-1.7b-instruct"
   icon: https://huggingface.co/datasets/HuggingFaceTB/images/resolve/main/banner_smol.png
@@ -651,9 +667,9 @@
     - https://huggingface.co/leafspark/Reflection-Llama-3.1-70B-bf16
     - https://huggingface.co/senseable/Reflection-Llama-3.1-70B-gguf
   description: |
-      Reflection Llama-3.1 70B is (currently) the world's top open-source LLM, trained with a new technique called Reflection-Tuning that teaches a LLM to detect mistakes in its reasoning and correct course.
+    Reflection Llama-3.1 70B is (currently) the world's top open-source LLM, trained with a new technique called Reflection-Tuning that teaches a LLM to detect mistakes in its reasoning and correct course.
 
-      The model was trained on synthetic data generated by Glaive. If you're training a model, Glaive is incredible — use them.
+    The model was trained on synthetic data generated by Glaive. If you're training a model, Glaive is incredible — use them.
   overrides:
     parameters:
       model: Reflection-Llama-3.1-70B-q4_k_m.gguf
@@ -973,15 +989,15 @@
     - https://huggingface.co/Sao10K/L3.1-8B-Niitama-v1.1
     - https://huggingface.co/Lewdiculous/L3.1-8B-Niitama-v1.1-GGUF-IQ-Imatrix
   description: |
-   GGUF-IQ-Imatrix quants for Sao10K/L3.1-8B-Niitama-v1.1
-   Here's the subjectively superior L3 version: L3-8B-Niitama-v1
-   An experimental model using experimental methods.
+    GGUF-IQ-Imatrix quants for Sao10K/L3.1-8B-Niitama-v1.1
+    Here's the subjectively superior L3 version: L3-8B-Niitama-v1
+    An experimental model using experimental methods.
 
-   More detail on it:
+    More detail on it:
 
-   Tamamo and Niitama are made from the same data. Literally. The only thing that's changed is how theyre shuffled and formatted. Yet, I get wildly different results.
+    Tamamo and Niitama are made from the same data. Literally. The only thing that's changed is how theyre shuffled and formatted. Yet, I get wildly different results.
 
-   Interesting, eh? Feels kinda not as good compared to the l3 version, but it's aight.
+    Interesting, eh? Feels kinda not as good compared to the l3 version, but it's aight.
   overrides:
     parameters:
       model: L3.1-8B-Niitama-v1.1-Q4_K_M-imat.gguf
@@ -1606,15 +1622,36 @@
   urls:
     - https://huggingface.co/Lewdiculous/MN-12B-Lyra-v4-GGUF-IQ-Imatrix
   description: |
-      A finetune of Mistral Nemo by Sao10K.
-      Uses the ChatML prompt format.
+    A finetune of Mistral Nemo by Sao10K.
+    Uses the ChatML prompt format.
   overrides:
     parameters:
       model: MN-12B-Lyra-v4-Q4_K_M-imat.gguf
   files:
     - filename: MN-12B-Lyra-v4-Q4_K_M-imat.gguf
       sha256: 1989123481ca1936c8a2cbe278ff5d1d2b0ae63dbdc838bb36a6d7547b8087b3
       uri: huggingface://Lewdiculous/MN-12B-Lyra-v4-GGUF-IQ-Imatrix/MN-12B-Lyra-v4-Q4_K_M-imat.gguf
+- !!merge <<: *mistral03
+  name: "magnusintellectus-12b-v1-i1"
+  url: "github:mudler/LocalAI/gallery/chatml.yaml@master"
+  icon: https://cdn-uploads.huggingface.co/production/uploads/66b564058d9afb7a9d5607d5/hUVJI1Qa4tCMrZWMgYkoD.png
+  urls:
+    - https://huggingface.co/GalrionSoftworks/MagnusIntellectus-12B-v1
+    - https://huggingface.co/mradermacher/MagnusIntellectus-12B-v1-i1-GGUF
+  description: |
+    How pleasant, the rocks appear to have made a decent conglomerate. A-.
+
+    MagnusIntellectus is a merge of the following models using LazyMergekit:
+
+        UsernameJustAnother/Nemo-12B-Marlin-v5
+        anthracite-org/magnum-12b-v2
+  overrides:
+    parameters:
+      model: MagnusIntellectus-12B-v1.i1-Q4_K_M.gguf
+  files:
+    - filename: MagnusIntellectus-12B-v1.i1-Q4_K_M.gguf
+      sha256: c97107983b4edc5b6f2a592d227ca2dd4196e2af3d3bc0fe6b7a8954a1fb5870
+      uri: huggingface://mradermacher/MagnusIntellectus-12B-v1-i1-GGUF/MagnusIntellectus-12B-v1.i1-Q4_K_M.gguf
 - &mudler
   ### START mudler's LocalAI specific-models
   url: "github:mudler/LocalAI/gallery/mudler.yaml@master"
@@ -2134,7 +2171,7 @@
     - https://huggingface.co/EpistemeAI/Athena-codegemma-2-2b-it
     - https://huggingface.co/mradermacher/Athena-codegemma-2-2b-it-GGUF
   description: |
-      Supervised fine tuned (sft unsloth) for coding with EpistemeAI coding dataset.
+    Supervised fine tuned (sft unsloth) for coding with EpistemeAI coding dataset.
   overrides:
     parameters:
       model: Athena-codegemma-2-2b-it.Q4_K_M.gguf