Update README.md

README.md

@@ -1,8 +1,7 @@
 ---
-base_model:
-- Undi95/Meta-Llama-3-8B-Instruct-hf
 language:
 - en
+- ko
 pipeline_tag: text-generation
 tags:
 - mergekit
@@ -17,7 +16,6 @@ license_name: llama3
 license_link: LICENSE
 extra_gated_prompt: >-
   ### META LLAMA 3 COMMUNITY LICENSE AGREEMENT
-
   Meta Llama 3 Version Release Date: April 18, 2024
   
   "Agreement" means the terms and conditions for use, reproduction, distribution and modification of the
@@ -190,15 +188,10 @@ extra_gated_fields:
 extra_gated_description: The information you provide will be collected, stored, processed and shared in accordance with the [Meta Privacy Policy](https://www.facebook.com/privacy/policy/).
 extra_gated_button_content: Submit
 ---
-
-# Meta-Llama-3-11.5B-Instruct
+# KoDolphin
 
 This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
 
-I had this idea at night that it would make sense to make a frankenmerge of Llama 3.. since we didn't get a 13B or 34B versions this time..
-
-Here's the same thing but for the base model: [mpasila/Meta-Llama-3-11.5B](https://huggingface.co/mpasila/Meta-Llama-3-11.5B/)
-
 ## Merge Details
 ### Merge Method
 
@@ -207,7 +200,8 @@ This model was merged using the passthrough merge method.
 ### Models Merged
 
 The following models were included in the merge:
-* [Undi95/Meta-Llama-3-8B-Instruct-hf](https://huggingface.co/Undi95/Meta-Llama-3-8B-Instruct-hf)
+* [beomi/Llama-3-Open-Ko-8B-Instruct-preview](https://huggingface.co/beomi/Llama-3-Open-Ko-8B-Instruct-preview)
+* [cognitivecomputations/dolphin-2.9-llama3-8b](https://huggingface.co/cognitivecomputations/dolphin-2.9-llama3-8b)
 
 ### Configuration
 
@@ -216,12 +210,14 @@ The following YAML configuration was used to produce this model:
 ```yaml
 slices:
   - sources:
-    - model: Undi95/Meta-Llama-3-8B-Instruct-hf
-      layer_range: [0, 24]
+      - model: beomi/Llama-3-Open-Ko-8B-Instruct-preview
+        layer_range: [0, 20]  # Use foundational and intermediate language processing layers in Korean
   - sources:
-    - model: Undi95/Meta-Llama-3-8B-Instruct-hf
-      layer_range: [8, 32]
-merge_method: passthrough
-dtype: bfloat16
+      - model: cognitivecomputations/dolphin-2.9-llama3-8b
+        layer_range: [15, 24]  # Utilize advanced coding and domain-specific layers
+
+merge_method: passthrough  # Direct combination of layers without transformation
+dtype: float16  # Efficient resource usage
+
 
-```
+```