Hugging Face's logo

KIEFERSA
/
Sophea-Nemo-Reranker

Text Ranking
Transformers
Safetensors
Greek
English
llama_bidirec
text-classification
greek
english
reranker
cross-encoder
rag
nemotron
custom_code
Eval Results (legacy)
text-embeddings-inference
Model card Files
xet
 Community
Sophea-Nemo-Reranker
File size: 25,077 Bytes
c58d757
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
---
language:
- el
- en
license: other
license_name: nvidia-open-model-license
base_model: nvidia/llama-nemotron-rerank-1b-v2
pipeline_tag: text-ranking
library_name: transformers
tags:
- greek
- english
- reranker
- cross-encoder
- rag
- nemotron
model-index:
- name: Sophea-Nemo-Reranker
  results:
  - task:
      type: text-ranking
      name: Greek+English reranking (5 domains, rerank top-50)
    dataset:
      type: retrieval
      name: Internal Greek+English retrieval benchmark
    metrics:
    - type: nDCG@10
      name: nDCG@10 (mean of 5 domains)
      value: 0.8172
    - type: recall@10
      name: Recall@10 (mean of 5 domains)
      value: 0.9267
---

<div style="display:flex;align-items:center;gap:18px;padding:16px 20px;margin-bottom:16px;border-radius:14px;background:linear-gradient(100deg,#e8f5ec 0%,#f7fbf8 62%);border:1px solid #c6e7d1"><img alt="KIEFERSA" style="flex:0 0 auto;height:58px;width:58px;border-radius:12px;object-fit:contain" src="data:image/webp;base64,UklGRgQNAABXRUJQVlA4WAoAAAAQAAAAxwAAxwAAQUxQSIsKAAABsMf/nyE52q+qZjd2nluEh00OsY2zbdsOH51t27Zt2744F1vrrqrf69X1m+mumq4K/ooISG4bSZKUmIIOeag1w1Ud8wFwPDgH2Om9zYFD2AcXDIa/hdWdgAff+r6oNK7pDDzw1uPxRlQRVoccFlu3e+pj0zrksByD8huqjSFSAm0c2l+8ilrIEQxaT1qC2ljAYYJD89PnxaYQQ05spcdMp5ZEcI0f/CtqGVsygTXY4+vYJGLI4QJg7CeIxkIOFwyGvE0tLeG0Xs8rVLGFHBZb1aORCXAYcniOQdfb61FJjXaEcEb5NeuoWRI+63DhShrgrAmdtZy4iAa4kJPj0PSMOdRCjuCQO3oaDXAhR3Dg+/9Ko0/IERxgl29pDA05QgCM+ohGn7DDYPDr1EJPr2cUiT7hp5aEhvUBZMZ6AY3rC3BjkLVdga0PkFjTGUqCDykOMw9kwEVwKIL9c1zX1d/saIJUsCFvE4sntILyFYjvjgQmeFBwbCsuaAMCyldhhOqVAYERp7bm2s3MKTMoKTF6apugiEOrvaMr+XbADJq8JtXf391IQImt4aEqYpBH3kmtuaXS7E4gkRL1s73pBS8IEY1rLusYCHFi+PoQurwnkjdh+dQ2wHP+i71p/GgM/XaQggITFpzdIpaQoWL7ZlcaGtKD5LPnrBOaeB8tbO3XAzg5LylIsYe/HyY8FwuLNE4/uoSck5SkkO/38jsk2djc05uSK5uSlOvQZzsAcBESdGyLJrckP4hZkFLeGw7AuZ+kDnAXt6fVyIa0ol7u62u0SGfrri2nM5xDQpJ8oqefksbq7+hOk4Eddgmw4e7uwHLMa6TC6NEe9DLaYyfV13sY95JMv5AX4FxgK6sv7ZiFCVwkDFdIifj2EBp93GAvS6a29ivuFbRPx1Jzi135nndWsyKXMdZvxLDCo7kDlNT43Z5AjjYTFDiW6ceVFFEYNJmOqtCthdgLuCXSBLiDaIBzi5sJv5PjKRalf6PU+UNptZ0lKrZpx5WSo8wA6eLeHhBLkZiWsNGWG/P8nEmfnwySd499Og5AsMyDSye3ISuzexyKQnx7KGSfSzvS6FMEHIt+ui2wjEPrffFxUcaGg8g09IHJPqg1DgGe9Q34AeLQTSF0Wr5egOfMJlKBh5S4QQ+f9zNqqfwkS9bzMV3TCU4yHXsq1LDYut9Ti7i6G0Dzc0gHpX9kxcpvXIeqEas7QQmDNv9bhipSgYUE2PaXrUIVadJHwUhv7pWrUUU6oJAvHi0nLzUnK78fhJ7STrdWo4rCSWzNzphvFpSkvhYuGGxxv+ldDyM5DqXHzEAtdZr+HBZLr6ckShk+BAdx0O+mTCX1GSUUslcwFk8oprE9v0EtpU2/lBAAI99DLWW44AJg/EfmzKTr+0qSnb4y0SJMCMFg2FsapXTRvyY48L1/MhPCA4+t9wsKpXTVhyc4iCOmmZAUFlhsVY81oops+glTSZNTZpv1NBzw2LreVYdKard9kSyWFhMWmLgXAmg+KLtuHSpZhP5OJhi0vWC5CUn+w3Ic2l9IAhzaYXfyr1pjdttzBIfWE5bQ/OwAiwldbiMPQyZxGeBO+we1MQdYylYPNpoJGcSZ5Y6ZRs0BDhao7Z5VqKSPCA7soLz3fUc4kIFvIErpG5wD7JYX4LIB/WFw9IdGMkRfKM1ZjPxgNuZj+tOCS9xck12/NtckM6T+HSwh8r9F7z23uHpGDjTPiMwCiPjQ1TdcbzGM9XqOrgWOcSe5o/5CFRURdyNvbZYa3eO0hsxGHJIFZGQ1utxeRz+2uMettJywbCSIjG9WUssIViGpvD2wjKONFQfXAgCQdRB9AICxzIO+ALAhy7ouwNYHSB1vSsKKRSvuzsAEDx8OOgWnbr38jQFG/CC7tvaazaB8BTY+uV0RGkkg43nIsdXe0RkEVKzECOvv28JMCMbGsTU8vCUwIaB8FfkAUn1LJbAc84mbJ/33PxbD2eIoUT5rbmGe10dBH9ErOwITzBM02vYTOjKF+PpQc/sW7AchE5b9p7VLyfj7tKMZ+CFpxU3ua9GRxoVnN3fXHpzx7xsuTOPXuwFpnEyETNA4+/gmwEXoULH9sj8n5yeBhD3745CSWHzHQQvzkSXAcwCJJO7dD3uzWEKFMq24pzUl5yUVifLpjgBChAgd25KJLcnVTUWqNejdkaYgBwZSlVZc0o7U15SkrGwv9TfiK+ltzfVltCKlJW11i57sacRPUge4O7tScwpipLD+/m5mh4OAVBg90oNeTlvSnvrqGyqM+I6WCvWzfelN6Yo0O77mso4muHqNlIhvDqXRxwU2E5ZNbm1CkrdIqfGTsdTcYFcq5p3ZHLjwExXbd7vRIuoK29I9/YQmsfiHijT+diAnR+wKF/LrISIWbyh03Y7LkaMtOkn31V4mJHlC3oow70wa4IpOitXq0/EAnHsBscVTWpF1zS3u5O2hRjwg0rjq4va0YjrGoegX+5jDyv6m+vpKenjOcSgK5WNVkH3u7kZrvnPcf6q6vSWwjEPLbVFwLsNAZAJbyz6oNQ4BnvmNHyAO3RRCp+XrBXgOylZiowo8pFBvddMlc1BL7SdZsorra+sqm/5vuZFQw3IcOly2EnFNV4COl5Om2gBDvni0nLIEVSNWd4ISBpW31hjJNn+j1PlDaRWTDWt21jyTbEgfBeOCQff761FGOqib2EqOm2XyWaF+ECPbPtWISnpF8Y0f+DtqqQr1tRRoqn1Vo5TucC7de/YoPEozYHt8h1qqpP4cKgCj3sVYHOH14AJg3MeopUzbZyQEwM5foJYqk/CkUWwb/qa5zjb9UoID3+dH1FKGa8MFg74vmOfOtu9LcMgd9reZECZYbD0fbUQVpe1fS5QmJ80166gd/lrXu2tNhU3Zh5cqYzQ/b6GRoEC+eJRdvxaV1O76CZlg0OaC5SZaeEyqi9ruEtJQbNUXmVI6Xk3iXiAQDFpNWYramAUWt08nGvcCQDyj2anz6NuANaknVD3QgCryHcGh9OgZtJxaYlkkej2tUEmfEQLEAb/QUOAASxn0GqKUviI4wO7f0EBQLAqHpFEfGfERYQJc3jEXHwATwnb+2lwL3+CCweA3dGwai0tS3Nv/F1OQC+CJ9Xpe0hcYd7gRceTfRnyBxbblIzTAOcRd7Why2hwzgeDBjM531JHo4xSXnxNaTFiEKvIABmXXVNNPR25xm63aX7AccUjmaX/BSmqucZ51r6kbASLjzKMBrhi4lqoyYBkHpUbMOgCMzMg4CtEHADjLPOgLABuyrOu0fkDpmi7rAUi2WfgvYIGHvNnV3do5SH/gfEbjw1saCxGOZ6hntzWZM+xIifqNwdTChEPDD8cA8NgCDnnL/nJXaiFHxfbL/oy0f4YMNzbtqFJiIUdFCv85tTmxgEPeTxdPaEUs5OhI4coL2pJ/WAghdrbu2jLyNSqMw8Zq7+pCLeSYAPdYFbWQIyXK5/rQ6BNypER8fQi1oJLKPhoPYCzkKKnxm92AFKrgkmi/HSiIBZgEm3F0KbGQoyKN805tSizk6EjhkqktSfQJNvkB7pJ2wIyFG2JrbyynoSHkmC9hd3WjFnYweqQnDQ2h59W+1IIPAABWUDggUgIAAHAdAJ0BKsgAyAA+MRiJRCIhiBJcEAGCWlu4MAAZsUGEUQU6Vr/HVACel73mNKbZii7/0o6X9Y+s6FoAm3srqdXIcWd/45kb7mdOviNuuq60W3qXx/nI9sR42ICkrOwkSyYty0cun8cMpqXF0fh8fW3dSwaV0akfevRFNfnTLgy9bQM6IEWlj9O95hHRhKjNbDCS4qWJnQbJto47spFiqntPEoRO945gp7f+FzLmeAJQakEjD+xgQDxMOYrHw79WEkLQkKplqTo/uTX4n54eEP0N4epeSNwNE7Cl7q0Xi32YrSRma3RieLfasZELkHnrVj60WkAA/uouoblbKpSNm/wu8g0tISmLMdPrSIKVi8oMaNvyXkZc6zskreB0cAODf99ubzUwj0N8IJywGgCWpByLV1Kx+G8TMEVJa0HKGlyheflfEABBIEOGipfrRMJEDIiV7tNyTCRkHRlacBBPJNrPF6gu5KqHk0mGh2xBZYDKEWXI1A8R9WyRig0s6CIELwgJcqvHXXixg1yTYYd3Ja2rm0FNGHOb8MdNErR6nH3DYmradAb7JnP/RNqQ3gvhl/pA7ygUm6CIWAO/JF0d93ZViT6KqEgvSVdzZIwTzPR1TQ5YDMPK1u42qqN2SqTRwtzHrtold7EeIrx7plafSjlNtoSSwuC99MOigEHtk+7uPfbMdQA6uO4+wAsQ6bfVGy+dpyM63pbfBdYa2IbgqA8ZA3EUsyDx6GwlVker0XVZhE/HSCyOiMAVqYYDYXpIECKSg6k4BD3OKMJ7nJlAAAAAAA=="/><div><div style="font-size:1.24rem;font-weight:700;color:#003713;line-height:1.2">Sophea-Nemo-Reranker</div><div style="font-size:.86rem;color:#3a5a47;margin-top:2px">Greek + English cross-encoder reranker — fine-tuned NVIDIA Llama-Nemotron-Rerank-1B-v2</div></div></div>

**Sophea-Nemo-Reranker** is a Greek + English **cross-encoder reranker**, fine-tuned from
**[NVIDIA llama-nemotron-rerank-1b-v2](https://huggingface.co/nvidia/llama-nemotron-rerank-1b-v2)** for
Greek and English reranking. It is the second (precision) stage of a
two-stage Greek RAG stack; the matching first stage is
**[Sophea-Nemo-Embedding](https://huggingface.co/KIEFERSA/Sophea-Nemo-Embedding)**.

- **Creator:** Kiefer SA
- **Base model:** `nvidia/llama-nemotron-rerank-1b-v2` (`LlamaBidirectionalForSequenceClassification`,
  **bidirectional**, single relevance logit; `trust_remote_code`)
- **Languages:** Greek + English
- **Input template:** a single sequence — `question:{q} \n \n passage:{p}`
- **Training:** BCE on labeled (query, doc) pairs, 1 epoch, max-len 4096, bf16

> **Greek is out-of-distribution for the base.** The base reranker lists 26 languages and **Greek is
> not one of them** — this fine-tune adapts its trained relevance head to Greek.

## Why it matters — base vs fine-tuned

Reranking the fixed top-50 candidates over 5 domains (nDCG@10), the fine-tune improves on the base
Nemotron reranker **on every domain**:

![Greek+English reranking — base vs fine-tuned](base_vs_ft.png)

<div style="overflow-x:auto">
<table style="width:100%;border-collapse:collapse;font-size:13px;font-variant-numeric:tabular-nums;font-family:-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,sans-serif">
<thead><tr>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:left;">metric</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Val</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Energy</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Legal</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Finance</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Medical</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Mean</th>
</tr></thead><tbody>
<tr><td style="padding:7px 8px;padding-left:20px;text-align:left;border-bottom:1px solid rgba(128,128,128,.15);">nDCG@10 — base</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.793</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.741</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.966</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.702</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.651</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.770</td></tr>
<tr><td style="padding:7px 8px;padding-left:20px;text-align:left;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">nDCG@10 — fine-tuned</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.874</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.780</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.969</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.754</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.709</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.817</td></tr>
<tr><td style="padding:7px 8px;padding-left:20px;text-align:left;border-bottom:1px solid rgba(128,128,128,.15);">Δ nDCG@10</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">+0.081</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">+0.039</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">+0.004</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">+0.052</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">+0.058</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">+0.047</td></tr>
<tr><td style="padding:7px 8px;padding-left:20px;text-align:left;border-bottom:1px solid rgba(128,128,128,.15);">Recall@10 — base</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.953</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.847</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.993</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.853</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.773</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.884</td></tr>
<tr><td style="padding:7px 8px;padding-left:20px;text-align:left;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">Recall@10 — fine-tuned</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.993</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.873</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.987</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.907</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.873</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.927</td></tr>
</tbody></table></div>

**Read.** Mean **nDCG@10 0.770 → 0.817
(+0.047)**, better on all five domains, with the
largest gains on the hardest ones (medical +0.06, finance +0.05, val +0.08). In the two-stage stack it
is also the strongest reranker we measured — edging a fine-tuned Qwen3-0.6B cross-encoder and beating
the base by a clear margin.

## Size comparison — vs base Qwen3-Reranker (off-the-shelf)

Reranking the same fixed top-50 candidates, same protocol (150 queries/domain), nDCG@10 — your
**fine-tuned 1B** reranker **edges the base Qwen3-Reranker-8B** and clearly beats 4B / 0.6B, at a
fraction of the size:

<div style="overflow-x:auto">
<table style="width:100%;border-collapse:collapse;font-size:13px;font-variant-numeric:tabular-nums;font-family:-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,sans-serif">
<thead><tr>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:left;">model</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Val</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Energy</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Legal</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Finance</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Medical</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Mean</th>
</tr></thead><tbody>
<tr><td style="padding:7px 8px;padding-left:20px;text-align:left;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">Sophea-Nemo-Reranker (ours · 1B · fine-tuned)</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.874</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.780</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.969</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.754</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.709</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.817</td></tr>
<tr><td style="padding:7px 8px;padding-left:20px;text-align:left;border-bottom:1px solid rgba(128,128,128,.15);">Qwen3-Reranker-8B (base, off-the-shelf)</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.858</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.796</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.975</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.763</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.672</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.813</td></tr>
<tr><td style="padding:7px 8px;padding-left:20px;text-align:left;border-bottom:1px solid rgba(128,128,128,.15);">Qwen3-Reranker-4B (base, off-the-shelf)</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.865</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.782</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.966</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.735</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.647</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.799</td></tr>
<tr><td style="padding:7px 8px;padding-left:20px;text-align:left;border-bottom:1px solid rgba(128,128,128,.15);">Qwen3-Reranker-0.6B (base, off-the-shelf)</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.812</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.746</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.960</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.692</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.634</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.769</td></tr>
</tbody></table></div>

**Read.** Mean nDCG@10 **0.817** (ours, 1B) vs **0.813**
(Qwen3-Reranker-8B), **0.799** (4B), **0.769** (0.6B) — and
ours is strongest on the hardest domain (medical 0.709 vs 8B's 0.672).
A small, Greek-tuned model beats a general reranker 8× its size.

## End-to-end two-stage pipeline

The two models are designed to run **together**: the embedder retrieves the top-50 candidates, the
reranker re-orders them. Measured end-to-end on the same 5-domain Greek+English eval (150 queries/set,
correct `query:`/`passage:` prompts):

![End-to-end two-stage stack — embedder alone vs + reranker](stack.png)

<div style="overflow-x:auto">
<table style="width:100%;border-collapse:collapse;font-size:13px;font-variant-numeric:tabular-nums;font-family:-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,sans-serif">
<thead><tr>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:left;">metric</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Val</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Energy</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Legal</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Finance</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Medical</th>
<th style="padding:10px 8px;font-weight:500;border-bottom:2px solid #0a8043;color:#0a8043;font-size:14px;text-align:center;">Mean</th>
</tr></thead><tbody>
<tr><td style="padding:7px 8px;padding-left:20px;text-align:left;border-bottom:1px solid rgba(128,128,128,.15);">nDCG@10 — embedder alone</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.863</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.802</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.939</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.702</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.689</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">0.799</td></tr>
<tr><td style="padding:7px 8px;padding-left:20px;text-align:left;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">nDCG@10 — + reranker (stack)</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.867</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.780</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.976</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.763</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.703</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.818</td></tr>
<tr><td style="padding:7px 8px;padding-left:20px;text-align:left;border-bottom:1px solid rgba(128,128,128,.15);">Δ from reranking</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">+0.004</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">-0.022</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">+0.037</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">+0.061</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">+0.013</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);">+0.019</td></tr>
<tr><td style="padding:7px 8px;padding-left:20px;text-align:left;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">Recall@10 — full stack</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.987</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.873</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.993</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.920</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.860</td><td style="padding:7px 8px;text-align:center;border-bottom:1px solid rgba(128,128,128,.15);font-weight:700;color:#0a8043;">0.927</td></tr>
</tbody></table></div>

**Read.** The full stack reaches **mean nDCG@10 0.818** (Recall@10 0.927) — the
strongest configuration measured. The reranker adds the most where the embedder is weakest
(**finance +0.061, legal +0.037**), i.e. it sharpens precision on the hard
cases. On **energy** it slightly regresses (-0.022) and on `val` the gain is marginal,
because the embedder alone is already very strong there and leaves little headroom — for the easiest,
already-high-recall domains you can skip the rerank stage if latency matters.


## Usage (transformers)

```python
import torch
from transformers import AutoTokenizer, AutoModelForSequenceClassification

name = "KIEFERSA/Sophea-Nemo-Reranker"
tok = AutoTokenizer.from_pretrained(name, trust_remote_code=True)
model = AutoModelForSequenceClassification.from_pretrained(
    name, num_labels=1, dtype=torch.bfloat16, trust_remote_code=True).eval()

def score(query, passage):
    text = f"question:{query} \n \n passage:{passage}"
    enc = tok(text, return_tensors="pt", truncation=True, max_length=4096)
    with torch.no_grad():
        return model(**enc).logits.squeeze(-1).item()   # higher = more relevant

# rank a candidate list by descending score
```

Feed the base's own `question:/passage:` template — it has a **trained** relevance head, so matching
its template adapts that head rather than fighting it. Needs `trust_remote_code=True`.

## Serving with vLLM

```bash
vllm serve KIEFERSA/Sophea-Nemo-Reranker --runner pooling --trust-remote-code
```

Exposes `/rerank`, `/v1/rerank` (Jina), `/v2/rerank` (Cohere) and `/score`:

```bash
curl http://localhost:8000/rerank -H 'Content-Type: application/json' -d '{
  "model": "KIEFERSA/Sophea-Nemo-Reranker",
  "query": "Ποια είναι η πρωτεύουσα της Ελλάδας;",
  "documents": ["Η Αθήνα είναι η πρωτεύουσα της Ελλάδας.", "Το Παρίσι είναι στη Γαλλία."]
}'
```

## License
Fine-tune of `nvidia/llama-nemotron-rerank-1b-v2` — governed by the **NVIDIA Open Model License** and
the **Llama 3.2 Community License**; commercial use permitted under those terms (note: **not**
Apache-2.0). Review the base model's license before use.