update

app/app.py

@@ -33,6 +33,7 @@ def retrieve_documents(occupation,skills):
     for german_label in skills:
         skill_query += german_label + ' '
     query = target_occupation_query + ' ' + skill_query
+    llama_query = 'info:' + target_occupation_name + ' ' + 'skills gap:' + skill_query
     print(query)
     docs = retriever.get_relevant_documents(query)
 
@@ -44,7 +45,7 @@ def retrieve_documents(occupation,skills):
     for doc in sorted_docs[:5]:
         doc_name = doc.metadata.get('name', 'Unnamed Document')
         doc_skill = doc.metadata.get('skills', '')
-        input_text = f"target occupation: {target_occupation_query}\n courses: name: {doc_name}, learning objectives: {doc_skill}"
+        input_text = f"target occupation: {llama_query}\n courses: name: {doc_name}, learning objectives: {doc_skill}"
         prompt = generate_prompt_exp(input_text)
         batch_prompts.append(prompt)
 

app/data/EduGBERT_cos_escoai/3b822eb4-09a8-4577-a432-58654a23254b/data_level0.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c8354fdb2c026dc2a76e8178aa3094ce31326c49e45ee29ffa050cc66a46ac4e
+oid sha256:932d15f4490ff30ab1d591f138d66086a4f10ef53e0d32fa6ed0bbae26081ea8
 size 29652000

app/data/EduGBERT_cos_escoai/3b822eb4-09a8-4577-a432-58654a23254b/index_metadata.pickle

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:113f97c64cd4e633ccd6a5a60a3cbc98b5eab33577ffe7ad4a8d89e044f3a165
+oid sha256:d050b5aede2ec58e0e741f33d0c4499c4b98dd200b5166b557d1a5ee49e2db23
 size 404064

app/data/EduGBERT_cos_escoai/3b822eb4-09a8-4577-a432-58654a23254b/length.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6c55a4f4498d15f6c63ec256c87fa63105b5ef2eb6100ff9df01dee8474b78e4
+oid sha256:faebe1fab42b5b6f39d3193b6a1396941178c28de5c11d0a3b2780232793cd8b
 size 28000

app/data/EduGBERT_cos_escoai/3b822eb4-09a8-4577-a432-58654a23254b/link_lists.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6cc81d4f04cfee68173d081839359f999b6c0c8595d339983318f951a9ebe73b
+oid sha256:012030771b78fb3011beddf18b990501672f8c5fb473ffb113f0e79ddd70d390
 size 59416

app/data/EduGBERT_cos_escoai/chroma.sqlite3

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:82a9eec36866175008a143ccd5be3e6cc72eec4b1a53aa2f5d0af5bd8e2b0615
-size 113496064
+oid sha256:151c1ef6c86444719fcd68cef527bab9e48ba631990b9296cd17c822c0011ca0
+size 113479680

app/embedding_setup.py

@@ -128,8 +128,8 @@ def evaluate(
 
 def compare_docs_with_context(doc_a, doc_b, target_occupation_name, target_occupation_dsp,skill_gap):
     
-    courses = f"First: name: {doc_a.metadata['name']}  description:{doc_a.metadata['description']} Second: name: {doc_b.metadata['name']}  description:{doc_b.metadata['description']}" 
-    #courses = f"First: name: {course_a['course_name']}  skills:{course_a['course_skills_edu']} Second: name: {course_b['course_name']}  skills:{course_b['course_skills_edu']}" 
+    #courses = f"First: name: {doc_a.metadata['name']}  description:{doc_a.metadata['description']} Second: name: {doc_b.metadata['name']}  description:{Sdoc_b.metadata['description']}" 
+    courses = f"First: name: {doc_a.metadata['name']}  learning outcomes:{doc_a.metadata['skills']} Second: name: {doc_b.metadata['name']}  learning outcomes:{doc_b.metadata['skills']}" 
     target_occupation = f"name: {target_occupation_name} description: {target_occupation_dsp}"
     skill_gap = skill_gap
     prompt = generate_prompt(target_occupation, skill_gap, courses)