NVIDIA Megatron-LM ensemble_classifer script code injection vulnerability

NVIDIA Megatron-LM is a PyTorch-based distributed training framework from NVIDIA that specializes in training large Transformer language models. A code injection vulnerability exists in the NVIDIA Megatron-LM ensembleclassifer script, which can be exploited by attackers to cause code execution,...

NVIDIA Megatron-LM msdp preprocessing script code injection vulnerability

NVIDIA Megatron-LM is a PyTorch-based distributed training framework from NVIDIA that specializes in training large Transformer language models. A code injection vulnerability exists in the NVIDIA Megatron-LM msdp preprocessing script, which can be exploited by attackers to cause code execution,...

CVE-2025-23353

NVIDIA Megatron-LM for all platforms contains a vulnerability in the msdp preprocessing script where malicious data created by an attacker may cause an injection. A successful exploit of this vulnerability may lead to code execution, escalation of privileges, Information disclosure, and data...

CVE-2025-23354

NVIDIA Megatron-LM for all platforms contains a vulnerability in the ensembleclassifer script where malicious data created by an attacker may cause an injection. A successful exploit of this vulnerability may lead to code execution, escalation of privileges, Information disclosure, and data...

CVE-2025-23349

NVIDIA Megatron-LM for all platforms contains a vulnerability in the tasks/orqa/unsupervised/nq.py component, where an attacker may cause a code injection. A successful exploit of this vulnerability may lead to code execution, escalation of privileges, information disclosure, and data tampering...

Arbitrary Code Injection

Overview megatron-core is a Megatron Core - a library for efficient and scalable training of transformer based models Affected versions of this package are vulnerable to Arbitrary Code Injection in the tasks/orqa/unsupervised/nq.py component. An attacker can execute arbitrary code, escalate...

collie-lm (>=1.0.5, <=1.0.7), frankenstein-model (>=5.1.6, <=5.3.9) +13 more potentially affected by CVE-2025-23349 via megatron-core (>=0.2.0, <=0.10.0)

megatron-core PYPI version =0.2.0, =1.0.5, =5.1.6, =0.1.0, =0.1.0, =1.0.0, =2.0.8, =1.0.0, =1.0.0, =1.0.0, =1.0.0, =1.0.0, =1.0.0, =1.4.9 Source cves: CVE-2025-23349 Source advisory: SNYK:PYTHON-MEGATRONCORE-13110025...

Arbitrary Code Injection

Overview megatron-core is a Megatron Core - a library for efficient and scalable training of transformer based models Affected versions of this package are vulnerable to Arbitrary Code Injection in the pretraingpt script. An attacker can execute arbitrary code, escalate privileges, access sensiti...

collie-lm (>=1.0.5, <=1.0.7), frankenstein-model (>=5.1.6, <=5.3.9) +13 more potentially affected by CVE-2025-23348 via megatron-core (>=0.2.0, <=0.10.0)

megatron-core PYPI version =0.2.0, =1.0.5, =5.1.6, =0.1.0, =0.1.0, =1.0.0, =2.0.8, =1.0.0, =1.0.0, =1.0.0, =1.0.0, =1.0.0, =1.0.0, =1.4.9 Source cves: CVE-2025-23348 Source advisory: SNYK:PYTHON-MEGATRONCORE-13110024...

Arbitrary Code Injection

Overview megatron-core is a Megatron Core - a library for efficient and scalable training of transformer based models Affected versions of this package are vulnerable to Arbitrary Code Injection via the msdp\preprocessing script. An attacker can execute arbitrary code and escalate privileges...

collie-lm (>=1.0.5, <=1.0.7), frankenstein-model (>=5.1.6, <=5.3.9) +13 more potentially affected by CVE-2025-23353 via megatron-core (>=0.2.0, <=0.10.0)

megatron-core PYPI version =0.2.0, =1.0.5, =5.1.6, =0.1.0, =0.1.0, =1.0.0, =2.0.8, =1.0.0, =1.0.0, =1.0.0, =1.0.0, =1.0.0, =1.0.0, =1.4.9 Source cves: CVE-2025-23353 Source advisory: SNYK:PYTHON-MEGATRONCORE-13110023...

collie-lm (>=1.0.5, <=1.0.7), frankenstein-model (>=5.1.6, <=5.3.9) +13 more potentially affected by CVE-2025-23354 via megatron-core (>=0.2.0, <=0.10.0)

megatron-core PYPI version =0.2.0, =1.0.5, =5.1.6, =0.1.0, =0.1.0, =1.0.0, =2.0.8, =1.0.0, =1.0.0, =1.0.0, =1.0.0, =1.0.0, =1.0.0, =1.4.9 Source cves: CVE-2025-23354 Source advisory: SNYK:PYTHON-MEGATRONCORE-13110088...

Arbitrary Code Injection

Overview megatron-core is a Megatron Core - a library for efficient and scalable training of transformer based models Affected versions of this package are vulnerable to Arbitrary Code Injection via the ensembleclassifer script. An attacker can execute arbitrary code, escalate privileges, disclos...

collie-lm (>=1.0.5, <=1.0.7), frankenstein-model (>=5.1.6, <=5.3.9) +13 more potentially affected by CVE-2025-23354 via megatron-core (>=0.2.0, <=0.10.0)

megatron-core PYPI version =0.2.0, =1.0.5, =5.1.6, =0.1.0, =0.1.0, =1.0.0, =2.0.8, =1.0.0, =1.0.0, =1.0.0, =1.0.0, =1.0.0, =1.0.0, =1.4.9 Source cves: CVE-2025-23354 Source advisory: MEND:CVE-2025-23354...

CVE-2025-23354

NVIDIA Megatron-LM for all platforms contains a vulnerability in the ensembleclassifer script where malicious data created by an attacker may cause an injection. A successful exploit of this vulnerability may lead to code execution, escalation of privileges, Information disclosure, and data...

CVE-2025-23354

NVIDIA Megatron-LM for all platforms contains a vulnerability in the ensembleclassifer script where malicious data created by an attacker may cause an injection. A successful exploit of this vulnerability may lead to code execution, escalation of privileges, Information disclosure, and data...

CVE-2025-23354

NVIDIA Megatron-LM for all platforms contains a vulnerability in the ensembleclassifer script where malicious data created by an attacker may cause an injection. A successful exploit of this vulnerability may lead to code execution, escalation of privileges, Information disclosure, and data...

CVE-2025-23354

NVIDIA Megatron-LM for all platforms contains a vulnerability in the ensembleclassifer script where malicious data created by an attacker may cause an injection. A successful exploit of this vulnerability may lead to code execution, escalation of privileges, Information disclosure, and data...

CVE-2025-23354

The CVE-2025-23354 issue affects NVIDIA Megatron-LM, specifically the ensemble_classifer script, with a code injection vulnerability that attacker-supplied data can trigger. The vulnerability may enable code execution, privilege escalation, information disclosure, and data tampering. Affected com...

CVE-2025-23353

NVIDIA Megatron-LM for all platforms contains a vulnerability in the msdp preprocessing script where malicious data created by an attacker may cause an injection. A successful exploit of this vulnerability may lead to code execution, escalation of privileges, Information disclosure, and data...