From ad001dae84858c97dd1a6dae5c550be506b5252a Mon Sep 17 00:00:00 2001
From: extio-de <github@extio.de>
Date: Sun, 29 Dec 2024 12:05:15 +0100
Subject: [PATCH] Misc improvements;

---
 src/main/java/de/extio/lmdatasetprep/Execution.java | 12 ++++++------
 src/main/resources/application.properties           |  2 +-
 src/main/resources/gemma.properties                 | 11 +++++++++++
 src/main/resources/llama3-cold.properties           |  2 +-
 src/main/resources/llama3-hot.properties            |  2 +-
 5 files changed, 20 insertions(+), 9 deletions(-)
 create mode 100755 src/main/resources/gemma.properties

diff --git a/src/main/java/de/extio/lmdatasetprep/Execution.java b/src/main/java/de/extio/lmdatasetprep/Execution.java
index 03c182a..ad53b98 100644
--- a/src/main/java/de/extio/lmdatasetprep/Execution.java
+++ b/src/main/java/de/extio/lmdatasetprep/Execution.java
@@ -161,7 +161,7 @@ public class Execution {
 	}
 	
 	public static void streamOut(final Path f, final String propertyPrefix, final Properties properties, final Consumer<OutputStream> consumer) {
-		final var streamFuncs = new ArrayList<OutStream>();
+		final var streamFuncs = new ArrayList<StreamFunc>();
 		for (int i = 0; i < 2; i++) {
 			final var destination = properties.getProperty(propertyPrefix + "." + i);
 			if (destination != null) {
@@ -197,8 +197,8 @@ public class Execution {
 		
 	}
 	
-	private static OutStream setupStreamOutToFile(final Path f, final String destination) {
-		return new OutStream(() -> {
+	private static StreamFunc setupStreamOutToFile(final Path f, final String destination) {
+		return new StreamFunc(() -> {
 			final var out = Path.of(destination).resolve(f);
 			if (Files.exists(out)) {
 				LOGGER.info("File already exists: {} ", out);
@@ -226,8 +226,8 @@ public class Execution {
 				});
 	}
 	
-	private static OutStream setupStreamOutToQueue(final Path f, final String destination) {
-		return new OutStream(() -> {
+	private static StreamFunc setupStreamOutToQueue(final Path f, final String destination) {
+		return new StreamFunc(() -> {
 			return new ByteArrayOutputStream();
 		},
 				(stream) -> {
@@ -239,7 +239,7 @@ public class Execution {
 	public record WorkPacket(Path file, String text) {
 	}
 	
-	private record OutStream(Supplier<OutputStream> open, Consumer<OutputStream> afterClose) {
+	private record StreamFunc(Supplier<OutputStream> open, Consumer<OutputStream> afterClose) {
 		
 	}
 }
diff --git a/src/main/resources/application.properties b/src/main/resources/application.properties
index e0ef2e9..9ba4467 100755
--- a/src/main/resources/application.properties
+++ b/src/main/resources/application.properties
@@ -4,7 +4,7 @@ logging.level.de.extio=info
 
 profile.model.s=llama3-cold
 profile.model.m=mistral-cold
-profile.model.l=chatml-cold
+profile.model.l=gemma
 profile.model.hot=llama3-hot
 profile.model.cold=alpaca-cold
 
diff --git a/src/main/resources/gemma.properties b/src/main/resources/gemma.properties
new file mode 100755
index 0000000..9d9cb60
--- /dev/null
+++ b/src/main/resources/gemma.properties
@@ -0,0 +1,11 @@
+prompts=gemma2
+tokenEncoding=
+maxTokens=2000
+maxContextLength=4000
+temperature=0.5
+topP=1.0
+maxContinuations=1
+modelProvider=OAI_TEXT_COMPLETION
+modelName=
+url=http://localhost:5001
+apiKey=
diff --git a/src/main/resources/llama3-cold.properties b/src/main/resources/llama3-cold.properties
index 866ebcf..e1a458c 100755
--- a/src/main/resources/llama3-cold.properties
+++ b/src/main/resources/llama3-cold.properties
@@ -2,7 +2,7 @@ prompts=llama3
 tokenEncoding=
 maxTokens=2000
 maxContextLength=4000
-temperature=0.2
+temperature=0.3
 topP=1.0
 maxContinuations=1
 modelProvider=OAI_TEXT_COMPLETION
diff --git a/src/main/resources/llama3-hot.properties b/src/main/resources/llama3-hot.properties
index 3670231..4f1186c 100755
--- a/src/main/resources/llama3-hot.properties
+++ b/src/main/resources/llama3-hot.properties
@@ -2,7 +2,7 @@ prompts=llama3
 tokenEncoding=
 maxTokens=2000
 maxContextLength=4000
-temperature=0.4
+temperature=0.7
 topP=1.0
 maxContinuations=1
 modelProvider=OAI_TEXT_COMPLETION