arxiv:2502.18293
Taneesh Gupta
gupta-tanish
·
AI & ML interests
Post-Training @MicrosoftResearch
Recent Activity
updated
a dataset
6 days ago
gupta-tanish/Qwen2.5-14B-Instruct-top-vs-bottom
updated
a dataset
6 days ago
gupta-tanish/Qwen2.5-32B-Instruct-top-vs-bottom
published
a dataset
6 days ago
gupta-tanish/Qwen2.5-32B-Instruct-top-vs-bottom
Organizations
None yet
models
26
gupta-tanish/llama-off-policy-qwq-10k-perturbation-iter1
Text Generation
•
8B
•
Updated
•
129
gupta-tanish/llama-3-8b-instruct-refa-budget_length-256-lamda-1.0-iteration2
Text Generation
•
8B
•
Updated
•
3
gupta-tanish/llama-3-8b-instruct-refa-budget_length-256-lamda-20.0-iteration1
Text Generation
•
8B
•
Updated
•
3
gupta-tanish/llama-3-8b-instruct-refa-lr-1e-6-beta10-gamma4-lambda-1.0-eos-increase-iteration2-lamda-0.1
Text Generation
•
8B
•
Updated
•
3
gupta-tanish/llama-3-8b-instruct-refa-lr-1e-6-beta10-gamma4-lambda-0.1-eos-increase-iteration2
Text Generation
•
8B
•
Updated
•
3
gupta-tanish/llama3-8b-instruct-refa-eos-increase-lamda-0.001-lr-1e-6-iteration1
Text Generation
•
8B
•
Updated
•
3
gupta-tanish/llama3-8b-instruct-refa-eos-increase-lamda-0.01-lr-1e-6-iteration1
Text Generation
•
8B
•
Updated
•
3
gupta-tanish/llama3-8b-instruct-refa-eos-increase-lamda-0.1-lr-1e-6-iteration1
Text Generation
•
8B
•
Updated
•
2
gupta-tanish/llama3-8b-instruct-refa-eos-increase-lamda-1.0-lr-1e-6-iteration1
Text Generation
•
8B
•
Updated
•
3
gupta-tanish/mistral-7b-instruct-refa-iteration2
Text Generation
•
7B
•
Updated
•
4
datasets
156
gupta-tanish/Qwen2.5-14B-Instruct-top-vs-bottom
Viewer
•
Updated
•
62.1k
•
122
gupta-tanish/Qwen2.5-32B-Instruct-top-vs-bottom
Viewer
•
Updated
•
62.1k
•
55
gupta-tanish/Llama3-8B-Instruct-16-Responses-16-bins-MPO
Viewer
•
Updated
•
60.8k
•
112
gupta-tanish/Qwen2.5-32B-Instruct-top2vsbottom2-selection
Viewer
•
Updated
•
62.1k
•
96
gupta-tanish/Qwen2.5-14B-Instruct-top2vsbottom2-selection
Viewer
•
Updated
•
62.1k
•
127
gupta-tanish/Ultrafeedback-llama3-8b-instruct-v0.2-on-policy-clean-8-binned-data
Viewer
•
Updated
•
60.8k
•
96
gupta-tanish/Ultrafeedback-llama3-8b-instruct-v0.2-on-policy-clean-4-binned-data
Viewer
•
Updated
•
60.8k
•
101
gupta-tanish/Ultrafeedback-llama3-8b-instruct-v0.2-on-policy-clean-2-binned-data
Viewer
•
Updated
•
60.8k
•
126
gupta-tanish/Ultrafeedback-Binarized-max-score-diff
Viewer
•
Updated
•
219k
•
99
gupta-tanish/QwQ-Long-CoT-30k-subset-Llama3.1-8B-dynamic-perturbation-regex-generation-max-margin-logp-10
Viewer
•
Updated
•
59k
•
104