✨ Introduce new feature(s): new benchmarks page

parakeet-nest · Oct 23, 2024 · 0accab1 · 0accab1
1 parent a60be7c
commit 0accab1
Show file tree

Hide file tree

Showing 4 changed files with 46 additions and 43 deletions.
diff --git a/README.md b/README.md
@@ -1,51 +1,53 @@
-# Awesome SMLs
+# Awesome SLMs
 
 <div align="center">
   <img width="50%" height="50%" src="female-borg.jpg" alt="Description of the image">
 </div>
 
 
 
-This is the list of the SMLs I use on my Raspberry Pi5 (8GB RAM) with Ollama
+This is the list of the SLMs I use on my Raspberry Pi5 (8GB RAM) with Ollama
 
 | Name | Size | Tag | Remark | Kind | URL | Good on Pi5 | Usable on Pi5 |
 | --- | --- | --- | --- | --- | --- | --- | --- |
-| CodeGemma 2b | 1.6GB | 2B | Fill-in-the-middle code completion | code | [Link](https://ollama.com/library/codegemma:2b) | ❌ | ✅ |
-| Gemma 2b | 1.7GB | 2B |  |  | [Link](https://ollama.com/library/gemma:2b) | ❌ | ✅ |
-| Gemma2 2b | 1.6GB | 2B |  |  | [Link](https://ollama.com/library/gemma2:2b) | ❌ | ✅ |
-| All-Minilm 22m | 46MB | 22M | Only Embeddings | embedding | [Link](https://ollama.com/library/all-minilm:22m) | ✅ | ✅ |
-| All-Minilm 33m | 67MB | 33M | Only Embeddings | embedding | [Link](https://ollama.com/library/all-minilm:33m) | ✅ | ✅ |
-| DeepSeek Coder 1.3b | 776MB | 1.3B | Trained on both 87% code and 13% natural language | code | [Link](https://ollama.com/library/deepseek-coder) | ✅ | ✅ |
-| TinyLlama 1.1b | 638MB | 1.1B |  |  | [Link](https://ollama.com/library/tinyllama) | ✅ | ✅ |
-| TinyDolphin 1.1b | 637MB | 1.1B |  |  | [Link](https://ollama.com/library/tinydolphin) | ✅ | ✅ |
-| Phi3 Mini | 2.4GB | 3B |  |  | [Link](https://ollama.com/library/phi3:mini) | ❌ | ✅ |
-| Phi3.5 | 2.2GB | 3B |  |  | [Link](https://ollama.com/library/phi3.5) | ❌ | ✅ |
-| Granite-code 3b | 2.0GB | 3B |  | code | [Link](https://ollama.com/library/granite-code) | ❌ | ✅ |
-| Qwen2.5 0.5b | 398MB | 0.5B |  | chat, tools | [Link](https://ollama.com/library/qwen2.5:0.5b) | ✅ | ✅ |
-| Qwen2.5 1.5b | 986MB | 1.5B |  | chat, tools | [Link](https://ollama.com/library/qwen2.5:1.5b) | ❌ | ✅ |
-| Qwen2.5 3b | 1.9GB | 3B |  | chat, tools | [Link](https://ollama.com/library/qwen2.5:3b) | ❌ | ✅ |
-| Qwen2.5 Coder 1.5b | 986MB | 1.5B |  | code, tools | [Link](https://ollama.com/library/qwen2.5-coder:1.5b) | ❌ | ✅ |
-| Qwen2 0.5b | 352MB | 0.5B |  |  | [Link](https://ollama.com/library/qwen2:0.5b) | ✅ | ✅ |
-| Qwen2 1.5b | 934MB | 1.5B |  |  | [Link](https://ollama.com/library/qwen2:1.5b) | ❌ | ✅ |
-| Qwen 0.5b | 395MB | 0.5B |  |  | [Link](https://ollama.com/library/qwen:0.5b) | ✅ | ✅ |
-| Qwen2 Math 1.5b | 935MB | 1.5B | Specialized math language model | math | [Link](https://ollama.com/library/qwen2-math:1.5b) | ❌ | ✅ |
-| StarCoder 1b | 726MB | 1B | Code generation model | code | [Link](https://ollama.com/library/starcoder:1b) | ✅ | ✅ |
-| StarCoder2 3b | 1.7GB | 3B |  | code | [Link](https://ollama.com/library/starcoder2:3b) | ❌ | ✅ |
-| Stable LM 2 1.6b | 983MB | 1.6B | LLM trained on multilingual data in English, Spanish, German, Italian, French, Portuguese, and Dutch. |  | [Link](https://ollama.com/library/stablelm2) | ✅ | ✅ |
-| Stable Code 3b | 1.6GB | 3B | Coding model | code | [Link](https://ollama.com/library/stable-code:3b) | ❌ | ✅ |
-| Replete-Coder Qwen2 1.5b | 1.9GB | 1.5B | Coding capabilities + non-coding data, fully cleaned and uncensored (mat+tool? to be tested) | code | [Link](https://ollama.com/rouge/replete-coder-qwen2-1.5b:Q8) | ✅ | ✅ |
-| Dolphin-Phi 2.7b | 1.6GB | 2.7B | uncensored |  | [Link](https://ollama.com/library/dolphin-phi:2.7b) | ❌ | ✅ |
-| Dolphin gemma2 2b | 1.6GB | 2B |  |  | [Link](https://ollama.com/CognitiveComputations/dolphin-gemma2:2b) | ❌ | ✅ |
+| codegemma:2b | 1.6GB | 2B | Fill-in-the-middle code completion | code | [Link](https://ollama.com/library/codegemma:2b) | ❌ | ✅ |
+| gemma:2b | 1.7GB | 2B |  | chat | [Link](https://ollama.com/library/gemma:2b) | ❌ | ✅ |
+| gemma2:2b | 1.6GB | 2B |  | chat | [Link](https://ollama.com/library/gemma2:2b) | ❌ | ✅ |
+| all-minilm:22m | 46MB | 22M | Only Embeddings | embedding | [Link](https://ollama.com/library/all-minilm:22m) | ✅ | ✅ |
+| all-minilm:33m | 67MB | 33M | Only Embeddings | embedding | [Link](https://ollama.com/library/all-minilm:33m) | ✅ | ✅ |
+| deepseek-coder:1.3b | 776MB | 1.3B | Trained on both 87% code and 13% natural language | code | [Link](https://ollama.com/library/deepseek-coder:1.3b) | ✅ | ✅ |
+| tinyllama | 638MB | 1.1B |  | chat | [Link](https://ollama.com/library/tinyllama) | ✅ | ✅ |
+| tinydolphin | 637MB | 1.1B |  | chat | [Link](https://ollama.com/library/tinydolphin) | ✅ | ✅ |
+| phi3:mini | 2.4GB | 3B |  | chat | [Link](https://ollama.com/library/phi3:mini) | ❌ | ✅ |
+| phi3.5 | 2.2GB | 3B |  | chat | [Link](https://ollama.com/library/phi3.5) | ❌ | ✅ |
+| granite-code:3b | 2.0GB | 3B |  | code | [Link](https://ollama.com/library/granite-code:3b) | ❌ | ✅ |
+| qwen2.5:0.5b | 398MB | 0.5B |  | chat, tools | [Link](https://ollama.com/library/qwen2.5:0.5b) | ✅ | ✅ |
+| qwen2.5:1.5b | 986MB | 1.5B |  | chat, tools | [Link](https://ollama.com/library/qwen2.5:1.5b) | ❌ | ✅ |
+| qwen2.5:3b | 1.9GB | 3B |  | chat, tools | [Link](https://ollama.com/library/qwen2.5:3b) | ❌ | ✅ |
+| qwen2.5-coder:1.5b | 986MB | 1.5B |  | code, tools | [Link](https://ollama.com/library/qwen2.5-coder:1.5b) | ❌ | ✅ |
+| qwen2:0.5b | 352MB | 0.5B |  | chat | [Link](https://ollama.com/library/qwen2:0.5b) | ✅ | ✅ |
+| qwen2:1.5b | 934MB | 1.5B |  | chat | [Link](https://ollama.com/library/qwen2:1.5b) | ❌ | ✅ |
+| qwen:0.5b | 395MB | 0.5B |  | chat | [Link](https://ollama.com/library/qwen:0.5b) | ✅ | ✅ |
+| qwen2-math:1.5b | 935MB | 1.5B | Specialized math language model | math | [Link](https://ollama.com/library/qwen2-math:1.5b) | ❌ | ✅ |
+| starcoder:1b | 726MB | 1B | Code generation model | code | [Link](https://ollama.com/library/starcoder:1b) | ✅ | ✅ |
+| starcoder2:3b | 1.7GB | 3B |  | code | [Link](https://ollama.com/library/starcoder2:3b) | ❌ | ✅ |
+| stablelm2:1.6b | 983MB | 1.6B | LLM trained on multilingual data in English, Spanish, German, Italian, French, Portuguese, and Dutch. | chat | [Link](https://ollama.com/library/stablelm2:1.6b) | ✅ | ✅ |
+| stable-code:3b | 1.6GB | 3B | Coding model | code | [Link](https://ollama.com/library/stable-code:3b) | ❌ | ✅ |
+| rouge/replete-coder-qwen2-1.5b:Q8 | 1.9GB | 1.5B | Coding capabilities + non-coding data, fully cleaned and uncensored (mat+tool? to be tested) | code | [Link](https://ollama.com/rouge/replete-coder-qwen2-1.5b:Q8) | ❌ | ✅ |
+| dolphin-phi:2.7b | 1.6GB | 2.7B | uncensored | chat | [Link](https://ollama.com/library/dolphin-phi:2.7b) | ❌ | ✅ |
+| CognitiveComputations/dolphin-gemma2:2b | 1.6GB | 2B |  | chat | [Link](https://ollama.com/CognitiveComputations/dolphin-gemma2:2b) | ❌ | ✅ |
 | allenporter/xlam:1b | 873MB | 1B |  | tools | [Link](https://ollama.com/allenporter/xlam:1b) | ❌ | ✅ |
 | sam4096/qwen2tools:0.5b | 352MB | 0.5B |  | tools | [Link](https://ollama.com/sam4096/qwen2tools:0.5b) | ✅ | ✅ |
 | sam4096/qwen2tools:1.5b | 935MB | 1.5B |  | tools | [Link](https://ollama.com/sam4096/qwen2tools:1.5b) | ❌ | ✅ |
-| mxbai-embed-large | 670MB | 335M | Only Embeddings | embedding | [Link](https://ollama.com/library/mxbai-embed-large:335m) | ✅ | ✅ |
-| nomic-embed-text | 274MB | 137M | Only Embeddings | embedding | [Link](https://ollama.com/library/nomic-embed-text:v1.5) | ✅ | ✅ |
-| Yi Coder 1.5b | 866MB | 1.5B | Code | code | [Link](https://ollama.com/library/yi-coder:1.5b) | ❌ | ✅ |
+| mxbai-embed-large:335m | 670MB | 335M | Only Embeddings | embedding | [Link](https://ollama.com/library/mxbai-embed-large:335m) | ✅ | ✅ |
+| nomic-embed-text:v1.5 | 274MB | 137M | Only Embeddings | embedding | [Link](https://ollama.com/library/nomic-embed-text:v1.5) | ✅ | ✅ |
+| yi-coder:1.5b | 866MB | 1.5B | Code | code | [Link](https://ollama.com/library/yi-coder:1.5b) | ❌ | ✅ |
 | bge-m3 | 1.2GB | 567M | Only Embeddings | embedding | [Link](https://ollama.com/library/bge-m3) | ❌ | ✅ |
-| reader-lm:0.5b | 352MB | 0.5b | convert HTML to Markdown |  | [Link](https://ollama.com/library/reader-lm:0.5b) | ✅ | ✅ |
-| reader-lm:1.5b | 935MB | 1.5b | convert HTML to Markdown |  | [Link](https://ollama.com/library/reader-lm:1.5b) | ✅ | ✅ |
-| shieldgemma:2b | 1.7GB | 2b | evaluate the safety of text |  | [Link](https://ollama.com/library/shieldgemma:2b) | ❌ | ✅ |
-| llama-guard3:1b | 1.6GB | 1b | evaluate the safety of text |  | [Link](https://ollama.com/library/llama-guard3:1b) | ❌ | ✅ |
+| reader-lm:0.5b | 352MB | 0.5b | convert HTML to Markdown | conversion | [Link](https://ollama.com/library/reader-lm:0.5b) | ✅ | ✅ |
+| reader-lm:1.5b | 935MB | 1.5b | convert HTML to Markdown | conversion | [Link](https://ollama.com/library/reader-lm:1.5b) | ✅ | ✅ |
+| shieldgemma:2b | 1.7GB | 2b | evaluate the safety of text | safety | [Link](https://ollama.com/library/shieldgemma:2b) | ❌ | ✅ |
+| llama-guard3:1b | 1.6GB | 1b | evaluate the safety of text | safety | [Link](https://ollama.com/library/llama-guard3:1b) | ❌ | ✅ |
 | granite3-dense:2b | 1.6GB | 2b |  | chat, tools, embedding | [Link](https://ollama.com/library/granite3-dense:2b) | ❌ | ✅ |
 | granite3-moe:1b | 822MB | 1b |  | chat, tools, embedding | [Link](https://ollama.com/library/granite3-moe:1b) | ✅ | ✅ |
+| llama3.2:1b | 1.3GB | 1b |  | chat, tools | [Link](https://ollama.com/library/llama3.2:1b) | ❌ | ✅ |
+| llama3.2:3b | 2.0GB | 3b |  | chat, tools | [Link](https://ollama.com/library/llama3.2:3b) | ❌ | ✅ |
diff --git a/benchmarks.html b/benchmarks.html
@@ -107,7 +107,7 @@
 
 
         <div class="header-section">
-            <h1>SMLs Benchmarks</h1>
+            <h1>SLMs Benchmarks</h1>
             <img src="male-borg.jpg" alt="AI Robot" class="header-image">
         </div>
 
@@ -116,7 +116,7 @@ <h1>SMLs Benchmarks</h1>
         </div>
 
         <p class="description">
-        <b>Chat completion:</b> benchmarks for different <b>SMLs</b> running on Ollama.<br>
+        <b>Chat completion:</b> benchmarks for different <b>SLMs</b> running on Ollama.<br>
     <pre><code>
     System instructions: You are useful AI agent.
     User question: Who is Jean-Luc Picard?

diff --git a/gen-readme.js b/gen-readme.js
@@ -38,8 +38,8 @@ function generateMarkdownTable(data, title, description, picture) {
   return markdown;
 }
 
-let title = "Awesome SMLs"
-let description = "This is the list of the SMLs I use on my Raspberry Pi5 (8GB RAM) with Ollama"
+let title = "Awesome SLMs"
+let description = "This is the list of the SLMs I use on my Raspberry Pi5 (8GB RAM) with Ollama"
 let picture = "female-borg.jpg"
 
 let mdContent = ``

diff --git a/index.html b/index.html
@@ -108,7 +108,7 @@
             text-align: center;
             margin-bottom: 2rem;
         }
-        
+
         .nav-links {
             margin: 1rem 0;
             padding: 1rem;
@@ -130,14 +130,15 @@
 <body>
     <div class="container">
         <div class="header-section">
-            <h1>Awesome SMLs</h1>
+            <h1>Awesome SLMs</h1>
             <img src="female-borg.jpg" alt="AI Robot" class="header-image">
         </div>
 
-        <p class="description">This is the list of the SMLs I use on my Raspberry Pi5 (8GB RAM) with Ollama</p>
+        <p class="description">This is the list of the SLMs I use on my Raspberry Pi5 (8GB RAM) with Ollama</p>
 
         <div class="nav-links">
-            📊 <a href="benchmarks.html">View Model Benchmarks</a>
+            📊 <a href="benchmarks.html">View Model Benchmarks</a><br>
+            📝 <a href="https://github.com/parakeet-nest/awesome-slms">Awesome SLMs on GitHub</a>
         </div>