From DeepSpeed to FSDP and Back Again with Hugging Face Accelerate
•
55
ContextualAI
company
Verified
AI & ML interests
Enterprise-grade retrieval augmented language models
Articles
Collections
3
Archangel is a suite of human feedback-aligned LLMs, released as part of the Human-Aware Loss Functions (HALOs) project by Ethayarajh et al. (2024).
Aligned models based on Zephyr-SFT from Table 2 and 3 in the KTO paper by Ethayarajh et al. (2024) (https://arxiv.org/pdf/2402.01306).
models
82
ContextualAI/Llama-200M
Text Generation
•
Updated
•
2.91k
ContextualAI/tiny-random-MistralForCausalLM
Text Generation
•
Updated
•
5.29k
ContextualAI/zephyr_sft_dpo
Text Generation
•
Updated
•
16
ContextualAI/zephyr_sft_kto_unary
Text Generation
•
Updated
•
18
ContextualAI/zephyr_sft_kto
Text Generation
•
Updated
•
16
•
1
ContextualAI/Contextual_KTO_Mistral_PairRM
Text Generation
•
Updated
•
45
•
31
ContextualAI/archangel_kto_pythia6-9b
Text Generation
•
Updated
•
18
ContextualAI/archangel_kto_pythia2-8b
Text Generation
•
Updated
•
44
ContextualAI/archangel_kto_pythia1-4b
Text Generation
•
Updated
•
21
ContextualAI/archangel_ppo_pythia2-8b
Text Generation
•
Updated
•
18
datasets
17
ContextualAI/LFQA_eval_dataset_unit_tests_contrastive
Viewer
•
Updated
•
260
•
35
ContextualAI/LFQA_eval_dataset_unit_tests_justification
Viewer
•
Updated
•
260
•
33
ContextualAI/LFQA_eval_dataset_unit_tests_with_justification
Viewer
•
Updated
•
5
•
36
ContextualAI/ultrafeedback_clair_32k
Viewer
•
Updated
•
29.1k
•
66
•
4
ContextualAI/ultrafeedback_rlaif-offpolicy_32k
Viewer
•
Updated
•
29.1k
•
43
•
1
ContextualAI/ultrafeedback_stronger-preferred_32k
Viewer
•
Updated
•
29.1k
•
43
•
1
ContextualAI/ultrafeedback_rlaif_32k
Viewer
•
Updated
•
31.8k
•
47
•
1
ContextualAI/ultrabin_clean_max_chosen_min_rejected_rationalized
Viewer
•
Updated
•
60.9k
•
17
ContextualAI/ultrabin_clean_max_chosen_rand_rejected_rationalized
Viewer
•
Updated
•
60.9k
•
8
ContextualAI/ultrabin_clean_max_chosen_min_rejected_rationalized_helpfulness
Viewer
•
Updated
•
60.9k
•
34