update README.md
Browse files
README.md
CHANGED
@@ -27,6 +27,8 @@ MiniCPM4 series are highly efficient large language models (LLMs) designed expli
|
|
27 |
- [MiniCPM4-0.5B](https://huggingface.co/openbmb/MiniCPM4-0.5B): The small version of MiniCPM4, with 0.5B parameters, trained on 1T tokens.
|
28 |
- [MiniCPM4-8B-Eagle-FRSpec](https://huggingface.co/openbmb/MiniCPM4-8B-Eagle-FRSpec): Eagle head for FRSpec, accelerating speculative inference for MiniCPM4-8B.
|
29 |
- [MiniCPM4-8B-Eagle-FRSpec-QAT](https://huggingface.co/openbmb/MiniCPM4-8B-Eagle-FRSpec-QAT): Eagle head trained with QAT for FRSpec, efficiently integrate speculation and quantization to achieve ultra acceleration for MiniCPM4-8B.
|
|
|
|
|
30 |
- [BitCPM4-0.5B](https://huggingface.co/openbmb/BitCPM4-0.5B): Extreme ternary quantization applied to MiniCPM4-0.5B compresses model parameters into ternary values, achieving a 90% reduction in bit width.
|
31 |
- [BitCPM4-1B](https://huggingface.co/openbmb/BitCPM4-1B): Extreme ternary quantization applied to MiniCPM3-1B compresses model parameters into ternary values, achieving a 90% reduction in bit width.
|
32 |
- [MiniCPM4-Survey](https://huggingface.co/openbmb/MiniCPM4-Survey): Based on MiniCPM4-8B, accepts users' quiries as input and autonomously generate trustworthy, long-form survey papers.
|
@@ -46,7 +48,6 @@ As of now, MiniCPM4-MCP supports the following:
|
|
46 |
|
47 |
- Cross-tool-calling capability: It can perform single- or multi-step tool calls using different tools that complies with the MCP.
|
48 |
|
49 |
-
|
50 |
## Evaluation
|
51 |
The detailed evaluation script can be found on the [GitHub](https://github.com/OpenBMB/MiniCPM/tree/minicpm-4/demo/minicpm4/MCP) page. The evaluation results are presented below.
|
52 |
|
@@ -71,8 +72,6 @@ The detailed evaluation script can be found on the [GitHub](https://github.com/O
|
|
71 |
| Whisper | 90.0 | 90.0 | 90.0 | 90.0 | 90.0 | 90.0 | 90.0 | 90.0 | 30.0 |
|
72 |
| **Average** | **80.2** | **70.2** | **49.1** | **83.5** | **67.7** | **43.8** | **88.3** | **76.1** | **51.2** |
|
73 |
|
74 |
-
|
75 |
-
|
76 |
## Statement
|
77 |
- As a language model, MiniCPM generates content by learning from a vast amount of text.
|
78 |
- However, it does not possess the ability to comprehend or express personal opinions or value judgments.
|
|
|
27 |
- [MiniCPM4-0.5B](https://huggingface.co/openbmb/MiniCPM4-0.5B): The small version of MiniCPM4, with 0.5B parameters, trained on 1T tokens.
|
28 |
- [MiniCPM4-8B-Eagle-FRSpec](https://huggingface.co/openbmb/MiniCPM4-8B-Eagle-FRSpec): Eagle head for FRSpec, accelerating speculative inference for MiniCPM4-8B.
|
29 |
- [MiniCPM4-8B-Eagle-FRSpec-QAT](https://huggingface.co/openbmb/MiniCPM4-8B-Eagle-FRSpec-QAT): Eagle head trained with QAT for FRSpec, efficiently integrate speculation and quantization to achieve ultra acceleration for MiniCPM4-8B.
|
30 |
+
- [MiniCPM4-8B-Eagle-vLLM](https://huggingface.co/openbmb/MiniCPM4-8B-Eagle-vLLM): Eagle head in vLLM format, accelerating speculative inference for MiniCPM4-8B.
|
31 |
+
- [MiniCPM4-8B-marlin-Eagle-vLLM](https://huggingface.co/openbmb/MiniCPM4-8B-marlin-Eagle-vLLM): Quantized Eagle head for vLLM format, accelerating speculative inference for MiniCPM4-8B.
|
32 |
- [BitCPM4-0.5B](https://huggingface.co/openbmb/BitCPM4-0.5B): Extreme ternary quantization applied to MiniCPM4-0.5B compresses model parameters into ternary values, achieving a 90% reduction in bit width.
|
33 |
- [BitCPM4-1B](https://huggingface.co/openbmb/BitCPM4-1B): Extreme ternary quantization applied to MiniCPM3-1B compresses model parameters into ternary values, achieving a 90% reduction in bit width.
|
34 |
- [MiniCPM4-Survey](https://huggingface.co/openbmb/MiniCPM4-Survey): Based on MiniCPM4-8B, accepts users' quiries as input and autonomously generate trustworthy, long-form survey papers.
|
|
|
48 |
|
49 |
- Cross-tool-calling capability: It can perform single- or multi-step tool calls using different tools that complies with the MCP.
|
50 |
|
|
|
51 |
## Evaluation
|
52 |
The detailed evaluation script can be found on the [GitHub](https://github.com/OpenBMB/MiniCPM/tree/minicpm-4/demo/minicpm4/MCP) page. The evaluation results are presented below.
|
53 |
|
|
|
72 |
| Whisper | 90.0 | 90.0 | 90.0 | 90.0 | 90.0 | 90.0 | 90.0 | 90.0 | 30.0 |
|
73 |
| **Average** | **80.2** | **70.2** | **49.1** | **83.5** | **67.7** | **43.8** | **88.3** | **76.1** | **51.2** |
|
74 |
|
|
|
|
|
75 |
## Statement
|
76 |
- As a language model, MiniCPM generates content by learning from a vast amount of text.
|
77 |
- However, it does not possess the ability to comprehend or express personal opinions or value judgments.
|