Update README.md

First GGUF conversion, added Airoboros 2.1 dataset. Originally a Meta Llama3, abliterated and retrained by by mlabonne, using black voodoo magic and a dorpo-dpo-mix-40 dataset to recover performance loss, caused by voodoo/abliteration.
There are gguf versions available, but I'd like to try a Imatrix (importance matrix) version, because they are supposed to run better on AMD gpu's using rocm. And I like it run on my previous gpu, an RX580 8GB, using the rocm fork of kobolcpp, cause, why not?
Added the Airoboros dataset from jondurbin for the same reason. Un-censoring was already achieved by the abliteration, but it wasn't trained for NSFW conversation. Supposedly, Airoboros includes that part, or something similar. It was a bit too long, so I didn't read the dataset. Or much else, I took the 2.1 dataset from Github because it has the word recommended behind in parenthesis and I'm a sucker for parenthesis, and everything advertised on HF as recommended.
No idea how it turned out, yet.

README.md

@@ -2,6 +2,7 @@
 base_model: mlabonne/NeuralDaredevil-8B-abliterated
 datasets:
 - mlabonne/orpo-dpo-mix-40k
+- jondurbin/airoboros-2.1
 license: llama3
 tags:
 - dpo
@@ -25,7 +26,8 @@ model-index:
       value: 69.28
       name: normalized accuracy
     source:
-      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Daredevil-8B-abliterated-dpomix
+      url: >-
+        https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Daredevil-8B-abliterated-dpomix
       name: Open LLM Leaderboard
   - task:
       type: text-generation
@@ -41,7 +43,8 @@ model-index:
       value: 85.05
       name: normalized accuracy
     source:
-      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Daredevil-8B-abliterated-dpomix
+      url: >-
+        https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Daredevil-8B-abliterated-dpomix
       name: Open LLM Leaderboard
   - task:
       type: text-generation
@@ -58,7 +61,8 @@ model-index:
       value: 69.1
       name: accuracy
     source:
-      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Daredevil-8B-abliterated-dpomix
+      url: >-
+        https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Daredevil-8B-abliterated-dpomix
       name: Open LLM Leaderboard
   - task:
       type: text-generation
@@ -74,7 +78,8 @@ model-index:
     - type: mc2
       value: 60
     source:
-      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Daredevil-8B-abliterated-dpomix
+      url: >-
+        https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Daredevil-8B-abliterated-dpomix
       name: Open LLM Leaderboard
   - task:
       type: text-generation
@@ -91,7 +96,8 @@ model-index:
       value: 78.69
       name: accuracy
     source:
-      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Daredevil-8B-abliterated-dpomix
+      url: >-
+        https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Daredevil-8B-abliterated-dpomix
       name: Open LLM Leaderboard
   - task:
       type: text-generation
@@ -108,7 +114,8 @@ model-index:
       value: 71.8
       name: accuracy
     source:
-      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Daredevil-8B-abliterated-dpomix
+      url: >-
+        https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Daredevil-8B-abliterated-dpomix
       name: Open LLM Leaderboard
 ---
 
@@ -154,4 +161,4 @@ Step 3: Run inference through the main binary.
 or 
 ```
 ./llama-server --hf-repo nielsijzerman/NeuralDaredevil-8B-abliterated-IQ4_NL-GGUF --hf-file neuraldaredevil-8b-abliterated-iq4_nl-imat.gguf -c 2048
-```
+```