Phi-4-Hindi

Phi-4-Hindi is a 14.7B multilingual large language model, instruction-tuned to achieve state-of-the-art performance in Hindi and English. Built on a pre-trained foundation, it is optimized for bilingual tasks with a diverse mixed-language dataset.

~1% better performance on English Tasks compared to the original (average benchmark scores)
~4% better performance on Hindi Tasks compared to the original (average benchmark scores)
~10% less emissions than the original (as reported on benchmark evaluations like open-llm-leaderboard)
Less Biases due to ordering of choices while answering MCQs

Model Details:

Developed by: Traversaal.ai, 1-800-LLMs
Language(s) (NLP): Optimized for Hindi and English
License: Apache 2.0
Paper : TBA April 15

Intended Use

We release Phi-4-Hindi under the Apache 2.0 license, encouraging researchers, developers, and enterprises to experiment with and build upon the model, particularly for bilingual, multilingual, and non-English applications. At the time of release, the model demonstrated state-of-the-art performance across an extensive English and Hindi evaluation suite.

Some potential downstream applications are as follows:

Research: This model serves as a valuable tool for researchers and developers working in NLP.
Commercial Use: It can be utilized as a foundational model for fine-tuning to meet specific industry needs.
Possible applications include:
- AI-powered Chat Assistants
- Customer Support Service
- Educational tools for language learning

Target audiences who may benefit from our model:

Academics: Researchers focused on Hindi and multilingual NLP advancements.
Businesses: Companies catering to Hindi-speaking and bilingual users.
Developers: Those integrating Hindi language capabilities into applications and services.
Educational Institutions: Schools and universities developing AI-powered learning tools.

Prompt Formats

Task	Input Format
Natural Language Inference	"`Text1 ### Text2 ### NLI ###`"
Multiple Choice Questions	"`Question ### A) a, B) b,... ### MCQ ###`"
Numeric Questions	"`Question ### NUMERIC ###`"
Boolean Questions	"`Question ### BOOLEAN ###`"
Questions seeking Long responses	"`Question ### LONG RESPONSE ###`"
Short responses (few words)	"`Input ### DIRECT RESPONSE ###`"
Coding	"`Input ### CODE ###`"
Text Summarization	"`Input ### SUMMARIZE ###`"
Paraphrasing/Rephrasing	"`Input ### PARAPHRASE ###`"
Translation to specified language	"`Input ### TRANSLATION [lang] ###`"
Text Simplification/ELI5	"`Input ### SIMPLIFY ###`"

The following prompt formats were used during training and are better suited for usage; however, the model works well even without such formatting.

Out-of-Scope Use

While Phi-4-Hindi is a powerful model designed for Hindi and English, usage must adhere to any applicable laws or regulations with its limitations in mind. The model must not be misused in ways that violate any applicable laws or regulations. Below are specific scenarios where its use is restricted:

Harmful or Malicious Use: The model should not be employed to create or distribute harmful, misleading, or inappropriate content, including but not limited to:
- Encouraging hate speech, violence, or discrimination
- Spreading misinformation or false narratives
- Facilitating or promoting illegal activities
Sensitive Data Handling: The model is not designed to process or generate personal, confidential, or sensitive information. Users must implement their own privacy policy and pipelines for handling sensitive contents.
Language Constraints: Phi-4-Hindi is a multilingual model built on the Phi-4 foundation but has been instruction-tuned for enhanced performance in Hindi and English. As a result, its proficiency in other languages may differ from the base Phi-4 model. The potential performance impact on other languages has not been evaluated, as this model is specifically optimized for bilingual (Hindi-English) tasks.
High-Risk Decision-Making: It should not be used for critical decision-making without human oversight, especially in medical, legal, financial, or safety-related contexts.

Bias, Risks, and Limitations

While efforts have been made to minimize biases, it is likely that the model, as with all LLM models, will exhibit some bias.

The model is trained as an AI assistant for Hindi and English speakers. The model is limited to producing responses for queries in these two languages and may not produce appropriate responses to other language queries.

By using this model, you acknowledge and accept that, as with any large language model, it may generate incorrect, misleading, and/or offensive information or content. The information is not intended as advice and should not be relied upon in any way, nor are we responsible for any of the content or consequences resulting from its use. We are continuously working to develop models with greater capabilities, and as such, welcome any feedback on the model.~~

Evaluation:

We evaluated our models on multiple well-known benchmarks to measure their effectiveness against other leading models, and the results are as follows:

Model	ARC-C	ARC-E	BoolQ	CMCQ	MMLU	Average*	MMLU-Pro	GPQA	MuSR	BBH	MATH-Hard
AryaBhatta-GemmaUltra-8.5B	22.70	25.04	22.95	62.23	23.70	31.32	22.66	25.34	42.72	41.12	2.95
Airavata-7B	25.09	30.47	25.31	62.17	33.20	35.25	16.35	27.43	37.57	36.00	13.60
sarvam-1-2B	30.03	33.25	62.17	42.80	27.90	39.23	-	-	-	-	-
Nemotron-4-Mini-Hindi-Instruct	55.80	71.63	62.11	68.10	43.20	60.17	25.95	30.87	41.53	40.11	2.04
Llama-3-Nanda-10B-Chat	65.36	80.64	82.29	67.60	50.61	69.30	31.57	30.12	43.52	49.38	5.59
Krutrim-2-12b-instruct	67.32	81.10	84.74	76.30	56.10	73.11	-	-	-	-	-
aya-expanse-8b	74.06	87.08	86.45	83.30	56.89	77.56	30.04	30.29	37.17	49.42	7.02
aya-expanse-32B	85.41	95.08	90.43	89.80	69.71	86.08	41.30	32.55	38.62	56.29	13.37
Our Qwen Model (14b)	90.61	94.82	88.53	90.70	75.00	87.93	52.63	36.24	44.84	64.97	25.08
Our Phi Model (14b)	97.39	92.24	87.65	87.40	75.59	88.05	52.39	39.77	49.07	66.97	23.11

Table 1: Metrics (.2f) of our models and other LLMs over several English benchmarks

Model	ARC-C	ARC-E	BoolQ	CMCQ	MMLU	Average
AryaBhatta-GemmaUltra-8.5B	22.70	25.08	22.95	62.17	23.80	31.34
Airavata-7B	22.87	25.13	23.28	62.17	33.20	33.33
sarvam-1-2B	32.76	35.06	62.16	47.10	24.22	40.26
Llama-3-Nanda-10B-Chat	45.99	60.56	71.96	54.70	36.35	53.91
Nemotron-4-Mini-Hindi-4B-Instruct	50.68	63.72	68.74	51.30	37.18	54.32
Krutrim-2-12b-instruct	56.83	70.66	78.86	64.10	46.51	63.39
aya-expanse-8b	57.42	72.90	80.42	69.00	43.39	64.63
aya-expanse-32B	73.29	85.48	87.73	79.70	56.96	76.63
Our Qwen Model (14b)	74.06	81.23	84.07	78.20	53.85	74.82
Our Phi Model (14b)	81.74	89.06	86.02	78.70	56.39	78.38

Table 2: Metrics (.2f) of our models and other LLMs over several Hindi benchmarks

Benchmark	Lang	Qwen-2.5-14B-Instruct	Our Qwen	Change	Phi-4	Our Phi	Change
ARC-Easy	En	95.45	94.82	🔻 0.63	97.31	97.39	🔼 0.08
	Hi	78.49	81.23	🔼 2.74	86.87	89.06	🔼 2.19
ARC-Challenge	En	90.87	90.61	🔻 0.26	92.41	92.24	🔻 0.17
	Hi	69.62	74.06	🔼 4.44	79.18	81.74	🔼 2.56
BoolQ	En	86.09	88.53	🔼 2.44	86.30	87.65	🔼 1.35
	Hi	78.89	84.07	🔼 5.18	82.72	86.02	🔼 3.30
Context-MCQ	En	91.20	90.70	🔻 0.50	86.30	87.40	🔼 1.10
	Hi	77.40	78.20	🔼 0.80	75.70	78.70	🔼 3.00
MMLU	En	74.37	75.00	🔼 0.63	74.67	75.59	🔼 0.92
	Hi	52.16	53.85	🔼 1.69	53.24	56.39	🔼 3.15
Average	En	87.60	87.93	🔼 0.33	87.40	88.05	🔼 0.65
	Hi	71.31	74.82	🔼 3.51	75.54	78.38	🔼 2.84
Overall		79.46	81.38	🔼 1.92	81.47	83.22	🔼 1.75

Table 3: Performance of our models compared to originals over each benchmark : evals through log likelihoods

Benchmark	Lang	Qwen-2.5-14B-Instruct	Our Qwen	Change	Phi-4	Our Phi	Change
MMLU-Pro	En	49.04	52.63	🔼 3.59	53.78	52.39	🔻 1.39
MATH hard	En	00.00	25.08	N/A	12.31	23.11	🔼 10.80
GPQA	En	32.21	36.24	🔼 4.03	33.72	39.77	🔼 6.05
MuSR	En	40.87	44.84	🔼 3.97	41.01	49.07	🔼 8.06
BigBench-Hard	En	63.74	64.97	🔼 1.23	68.60	66.97	🔻 1.63
Average		37.17	44.75	🔼 7.58	41.88	46.26	🔼 4.38

Table 4: Performance of our models compared to originals over each benchmark : evals through eval-harness

Recommendations

It is advisable for users to:

Refrain from deploying the model in sensitive domains without human supervision.
Cross-check factual information generated by the model for accuracy.
Continuously assess the model to ensure compliance with ethical standards.
Be mindful of potential biases and unintended outputs, especially in critical applications.

Emissions

We believe that our use of shorter, more compressed instruction-response pairs during training has led to a model that generates concise yet accurate responses. This approach allows the model to meet requirements and arrive at correct answers while also improving efficiency and reducing emissions.

Unlike distillation from reasoning or Chain-of-Thought (CoT) models, which often produce unnecessarily long responses—such as "Next, we proceed with..." or "Okay, let's do this..."—our method focuses purely on the essential step-by-step reasoning. For datasets containing multiple correct multi-step solutions, we prioritized training on the shortest valid solution, eliminating filler content.

This ensures that Phi-4-Hindi remains both effective and efficient, delivering high-quality results without unnecessary verbosity.

Model Responses vs Order of Choices in MCQs

Benchmarks like MMLU-Pro include up to 10 answer choices, whereas most training datasets typically contain only 4-5 choices. To improve robustness and reduce bias, we introduced modifications in the ordering and labeling of answer choices, including:

Reordering answer choices to create an imbalance that opposes the original model's choice distribution.
Altering labels in 5% of MCQ samples, replacing standard A/B/C/D labels with variations like a/b/c/d, 1/2/3/4, or w/x/y/z.

These adjustments resulted in reduced bias toward earlier answer choices, leading to a more balanced selection distribution compared to the original Phi-4. The images below illustrate the distribution of choices selected by the model during MMLU-Pro evaluation.

jebish7
/

For_readme