deepseek-ai/DeepSeek-V4-Pro
Text Generation • 862B • Updated • 5.4M • • 4.72k
Offer L.
OfferL
AI & ML interests
None yet
Recent Activity
liked a model 10 days ago
deepseek-ai/DeepSeek-V4-Pro liked a model 7 months ago
zai-org/GLM-4.6 liked a model 7 months ago
meta-llama/Llama-4-Scout-17B-16E-InstructOrganizations
liked a
model 10 days ago
liked 2
models 7 months ago
liked a
Space 8 months ago
liked a
Space 9 months ago
liked a
Space 11 months ago
reacted to chansung's post with 👍 over 1 year ago
Post
2067
Simple summarization of Evolving Deeper LLM Thinking (Google DeepMind)
The process starts by posing a question.
1) The LLM generates initial responses.
2) These generated responses are evaluated according to specific criteria (program-based checker).
3) The LLM critiques the evaluated results.
4) The LLM refines the responses based on the evaluation, critique, and original responses.
The refined response is then fed back into step 2). If it meets the criteria, the process ends. Otherwise, the algorithm generates more responses based on the refined ones (with some being discarded, some remaining, and some responses potentially being merged).
Through this process, it demonstrated excellent performance in complex scheduling problems (travel planning, meeting scheduling, etc.). It's a viable method for finding highly effective solutions in specific scenarios.
However, there are two major drawbacks:
🤔 An excessive number of API calls are required. (While the cost might not be very high, it leads to significant latency.)
🤔 The evaluator is program-based. (This limits its use as a general method. It could potentially be modified/implemented using LLM as Judge, but that would introduce additional API costs for evaluation.)
https://arxiv.org/abs/2501.09891
The process starts by posing a question.
1) The LLM generates initial responses.
2) These generated responses are evaluated according to specific criteria (program-based checker).
3) The LLM critiques the evaluated results.
4) The LLM refines the responses based on the evaluation, critique, and original responses.
The refined response is then fed back into step 2). If it meets the criteria, the process ends. Otherwise, the algorithm generates more responses based on the refined ones (with some being discarded, some remaining, and some responses potentially being merged).
Through this process, it demonstrated excellent performance in complex scheduling problems (travel planning, meeting scheduling, etc.). It's a viable method for finding highly effective solutions in specific scenarios.
However, there are two major drawbacks:
🤔 An excessive number of API calls are required. (While the cost might not be very high, it leads to significant latency.)
🤔 The evaluator is program-based. (This limits its use as a general method. It could potentially be modified/implemented using LLM as Judge, but that would introduce additional API costs for evaluation.)
https://arxiv.org/abs/2501.09891
liked a
model over 1 year ago
liked a
Space over 1 year ago
liked a
model over 1 year ago
liked a
model almost 2 years ago
liked 5
Spaces almost 2 years ago
Llama3.1 S V0.2 Checkpoint 2024 08 20
😻
115
Chat with Llama3.1 using spoken audio or synthesize speech
Kolors Virtual Try-On
👕
10.1k
Generate virtual try‑on images of clothes on a person
Falcon Mamba Playground
🐍
65
Generate chat responses using FalconMamba-7b model
Paligemma HF
🤗
85
Generate text and segment images using PaliGemma
FLUX.1 [dev]
🖥
9.47k
Generate images from text prompts
liked a
model almost 2 years ago
liked 3
Spaces almost 2 years ago